WikiDiscuss

WikiDiscuss


Magic Words

posts: 14214
Use this thread to discuss the Magic Words page.
posts: 14214

I just had a fucked up thought.

jcowan and I have been discussing that mex shouldn't have any internal
grammar; it should just be a list of words for symbols that the speakers can handle
however you like.

It just occured to me that the language already has a facility for that:
lo'u...le'u.

As an exmaple:

li mo'e lo'u pa su'i re le'u du li mo'e lo'u ci le'u

All three parsers accept this.

-Robin

posts: 14214

Why, exactly, is BAhE + BU invalid? The Red Book (C19S16 as usual) seems quite clear on this point:

``bu makes the preceding word into a lerfu word, except for ``zo, ``si, ``sa, ``su, ``lo'u, ZOI cmavo, ``fa'o, ``zei, BAhE cmavo, and ``bu. Multiple ``bu cmavo may be used in succession.

-Robin

posts: 14214

Does SU go back to the last NIhO, LU, TUhE, or TO (as grammar.300 claims) or the beginning of input (as the Red Book claims)?

-Robin

posts: 14214

grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU clauses. I'm pretty sure this is just a YACC side effect; it certainly seems amazingly ridiculous to me. If anyone disagrees, let me know.

I suppose I can imagine something like "mi zoi zoi quietude zoi zei cinmo", but I don't know if it's worthwhile building such a bizarre construction into the language. Maybe.

ZO-clause+ZEI and LOhU...LEhU+ZEI are just dumb as hell, though.

-Robin

posts: 14214

The red book claims that UI cmavo can't mark BU. What's up with that?

Specifically, it says:

UI and CAI cmavo mark the previous word, except for ``zo, ``si, ``sa, ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE cmavo, and ``bu. Multiple UI cmavo may be used in succession. A following ``nai'' is made part of the UI.

-Robin

posts: 14214

What do we do about SI, SA and ZOI?

Option 1:

  • SI is an allowed ZOI delimiter.
  • SI after ZOI eats through delimiters and text seperately, so four SI in a row are required to erase a ZOI.
  • Fastest way to erase a ZOI clause is to use SA + something that occured before the ZOI; SA+ZOI doesn't help at all; it just puts you back in the ZOI. Since SI can be a ZOI delimiter, SA+ZOI+SI is just a ZOI clause ready for non-Lojban text.


Option 2:

  • SI is an allowed ZOI delimiter.
  • SI after ZOI eats the entire thing, ZOI and all.
  • SA+ZOI puts you back in the ZOI. Since SI can be a ZOI delimiter, SA+ZOI+SI is just a ZOI clause ready for non-Lojban text.


Option 3:

  • SI is NOT an allowed ZOI delimiter.
  • SI after ZOI eats through delimiters and text seperately, so four SI in a row are required to erase a ZOI.
  • SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
  • This would mean that the only case where WORD + SA + same category as WORD + SI doesn't == nothing would be with ZO.


I have a strong preference for option three; it gives the most user friendliness, IMO. What do you all think?

-Robin

posts: 1912


Definition: A magic word is any cmavo of selmaho ZO, ZOI, LOhU, LEhU, ZEI, BU, SI, SA, SU, BAhE or FAhO
Definition: Y is not a word.

The exeptionless rules for magic words would be as follows:

ZO: Unless it has been turned off by a preceding magic word, it quotes the following word and, if it is a magic word, turns it off. If there is no following word, it gives an error.

ZOI: Unless it has been turned off by a preceding magic word, it turns the following word into a delimiter and, if it is a magic word, turns it off.
Everything that follows is non-lojban-text until the delimiter is found again (turned off if it's a magic word). If there is no following word, or if the following word does not reappear to close the quote, it gives an error.

LOhU: Unless it has been turned off by a preceding magic word, it quotes each following word until LEhU and, if it is a magic word, turns it off. If LEhU does not appear, it gives an error.

LEhU: Unless it has been turned off by a preceding magic word, it closes a LOhU quote. If it's not turned off and there is no opened LOhU it gives an error.

ZEI: Unless it has been turned off by a preceding magic word, it takes the previous word or magic word output, and joins it with the following word to make a tanru-unit. If the following word is a magic word, it turns it off. If there is no previous stuff, or if there is no following word, then it gives an error.

BU: Unless it has been turned off by a preceding magic word, it takes the previous word or magic word output, and turns it into a lerfu.

SI: Unless it has been turned off by a preceding magic word, it takes the previous word or magic word output and anihilates it. (i.e. now the previous stuff is the stuff before the one erased). If there is no previous stuff, it gives an error.

SA: Unless it has been turned off by a preceding magic word, it deletes everything back up to and including the first word found of the selmaho of the word that follows (stuff processed by preceding magic words in general does not keep its original selmaho) or else if no such word is found, up to the beginning of text. If no word follows, it gives an error.

SU: Unless it has been turned off by a preceding magic word, it deletes everything back up to the beginning of text. It never gives error.

BAhE: Unless it has been turned off by a preceding magic word, it modifies the word that follows, and becomes invisible (i.e. it is not part of the previous stuff for following magic words). If what follows is a magic word, it does not turn it off. If no word follows, it gives an error.

FAhO: Unless it has been turned off by a preceding magic word, it terminates the text. If the utterance so far is not a valid text, it gives an error.

Those rules are exeptionless, and they don't require to define any special order of precedence. The order is simply first come, first served. In some cases we may want, for whatever reason, to deviate from the exceptionless rules. For example, if we wanted {si} to erase {bu}, or {zoi}, or {zei}, or even {zo} for that matter. But the price of any exception is complication of the rules and opening a can of worms, because usually exceptions require more exceptions to handle special cases. I propose we start from the exceptionless rules and analyse carefully any proposed exception to them.

mu'o mi'e xorxes

posts: 1912


This are the answers to the Outstanding Questions, according to the exceptionless rules:

> Is nested lo'u...le'u allowed?

No, {lo'u} turns off every magic word that follows, including other {lo'u}s, up to the first {le'u} found.

> Not really, but lo'u has no effect and we allow "zo le'u".

The exceptionless rules do not allow {zo le'u} within the quote because {zo} is turned off and {le'u} then terminates it.

> Does zoi function in lo'u...le'u?
> No. If you need to quote a broken zoi, use another zoi.

Right.

> What does BAhE+BU do?
> Just empasizes the BU.

Right.

> The red book claims that UI cmavo can't mark BU. What's up with that?
> I have no idea, but I'm ignoring it. UI can mark anything that doesn't grab it.

Good. I have no idea what's up with that either.

> What do we do about ZOI, SI and SA interactions? Is SI allowed
> as a ZOI delimiter? Does SI after a ZOI clause erase the whole
> thing?

The exceptionless rules would accept SI and SA as delimiters.
SI after a ZOI-clause would erase the whole thing.

> SI, SA and SU are not allowed ZOI delimiters.
> 4 SI for ZOI erasure from outside.
> SA+ZOI+SI works. ZOI+SI == nothing.

Those would constitute exceptions to the exceptionless rules.

> Does SU go back to the last NIhO, LU, TUhE, or TO
> (as grammar.300 claims) or the beginning of input
> (as the Red Book claims)?
> The latter; SA can be used for the other things, and if SU works
> the former way there is no way to unequivocably erase to
> the beginning of input.

Good. Either way could work with the exceptionless rules, though.

> grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU
> clauses.

That agrees with the exceptionless rules. To clarify: ZEI would take a ZOI, ZO or LOhU-LEhU clause to its left, but just ZO, ZOI, LOhU or LEhU to its right. This is because ZEI cannot turn the magic off of those words once they have used it.

> I'm pretty sure this is just a YACC side effect; it certainly seems
>amazingly ridiculous to me. If anyone disagrees, let me know.
> It doesn't do any harm, and allows certain useful things (like ZEI
> lujvo with the various words that ZEI can't bind to in them).

And most importantly: It simplifies the rules.

mu'o mi'e xorxes

posts: 14214

On Mon, Nov 08, 2004 at 01:13:31PM -0800, Jorge Llamb?as wrote:
> --- Robin Lee Powell wrote:
> > > {broda ba'e si brode} would be an emphatically deleted
> > > {broda}, replaced by {brode}: i.e. {ba'e} is always (no
> > > exceptions) absorbed by the following word. The construct
> > > {ba'e si} would be an emphasized word eraser.
> >
> > I'm sorry, but no matter how many times I read that my brain
> > still treats it as an exception. SI is just *more* *important*
> > then BAhE to me, and it seems bizarre that SI wouldn't do its
> > proper job (erasing BAhE) in this case.
>
> If you say that SI has priority over everything, then it is an
> exception.

To *your* rules. Please try to remember that your rules are the new
kid on the block. :-)

> If you say that left-to-right processing has priority, then it is
> not an exception.

Oh, another thing: in left-to-right processing, BU *always* loses.
Not sure this is a bad thing, but wanted to point it out.

> But you don't consider SI more important than everything:

Never said I did.

> Having to remember the order of precedence of all the magic words,
> plus all the exceptions and counter-exceptions to that order of
> precedence, is too much trouble for what it buys us, in my
> opinion.

Whereas I have very little trouble with it, once they are all
written up and decided upon. Lots of programming experience, do ya
ken.

> > However, someone who "grew up" with your system likely wouldn't
> > feel that way. I just wanted to get it out.
>
> I don't know why you feel that SI is more important than BAhE but
> less important than ZO, other than because you have become used to
> it.

Isn't that exactly what I just said??

> I have not been able to learn the official precedence rules of
> magic words in about 10 years of doing Lojban. It is not even
> clear if there is an official set of consistent and complete
> rules.

There is most certainly *not*; this is what I'm trying to create.

> I just hope we don't end up with another unlearnable set.

My current rules are quite easy for me to learn, and as far as I
know they are consistent and complete.

Moving the learning issue to another post.

-Robin


posts: 1912


> Re: Magic Words
> Why, exactly, is BAhE + BU invalid? The Red Book (C19S16 as usual) seems
> quite clear on this point:
>
> ``bu makes the preceding word into a lerfu word, except for ``zo, ``si'',
> ``sa, ``su, ``lo'u, ZOI cmavo, ``fa'o, ``zei'', BAhE cmavo, and
> ``bu. Multiple ``bu cmavo may be used in succession.
>
> -Robin

It is not invalid, it is an emphasized {bu}.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 06, 2004 at 04:45:01PM -0800, Jorge Llamb?as wrote:
>
> --- wikidiscuss@lojban.org wrote:
> > Re: Magic Words
> > Why, exactly, is BAhE + BU invalid? The Red Book (C19S16 as
> > usual) seems quite clear on this point:
> >
> > ``bu'' makes the preceding word into a lerfu word, except for
> > ``zo, ``si, ``sa, ``su, ``lo'u, ZOI cmavo, ``fa'o,
> > ``zei, BAhE cmavo, and ``bu. Multiple ``bu'' cmavo may be
> > used in succession.
>
> It is not invalid, it is an emphasized {bu}.

Umm, OK, then what does the {bu} eat?

So you're saying that {broda ba'e bu} is a very lerfu broda?

That possibility really never occured to me, and seems amazingly
bizarre and pointless. I can imagine uses for a lerfu based on ba'e
or za'e (with some effort) but I cannot imagine why one would want
to emphasize or nonce-ize bu itself whilst still retaining its
grammatical function.

-Robin

--
http://www.digitalkingdom.org/~rlpowell/ *** http://www.lojban.org/
Reason #237 To Learn Lojban: "Homonyms: Their Grate!"
Proud Supporter of the Singularity Institute - http://singinst.org/


posts: 14214

On Sat, Nov 06, 2004 at 06:49:00PM -0800, wikidiscuss@lojban.org wrote:
> Re: Magic Words
> What do we do about SI, SA and ZOI?

Forgot some issues here:

> Option 3:
>
> * SI is NOT an allowed ZOI delimiter.
>
> * SI after ZOI eats through delimiters and text seperately, so
> four SI in a row are required to erase a ZOI.
>
> * SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI
> delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
>
> * This would mean that the only case where WORD + SA + same
> category as WORD + SI doesn't == nothing would be with ZO.

  • Furthermore, ZOI+SI == nothing, rather than being the beginning of

a ZOI clause that requires a 5th dan Lojban black belt to get out
of ("zoi .y. si .y. .y. si si si si si").

> I have a strong preference for option three; it gives the most
> user friendliness, IMO.

The new point I just added is one of the major reasons I prefer
this.

-Robin


posts: 381

In a message dated 2004-11-06 9:29:38 PM Eastern Standard Time, robin via
wikidiscuss@lojban.org writes:


> UI and CAI cmavo mark the previous word, except for ``zo, ``si, ``sa'',
> ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE cmavo, and ``bu''.

Looks like "uibu" is okay, but not "bu.ui".

stevo

posts: 14214

On Sat, Nov 06, 2004 at 10:14:38PM -0500,
MorphemeAddict@wmconnect.com wrote:
> In a message dated 2004-11-06 9:29:38 PM Eastern Standard Time,
> robin via wikidiscuss@lojban.org writes:
>
>
> > UI and CAI cmavo mark the previous word, except for ``zo'',
> > ``si, ``sa, ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE
> > cmavo, and ``bu''.
>
> Looks like "uibu" is okay, but not "bu.ui".

Well, yes. My question was, why?

-Robin


posts: 1912


> > Option 3:
> >
> > * SI is NOT an allowed ZOI delimiter.
> >
> > * SI after ZOI eats through delimiters and text seperately, so
> > four SI in a row are required to erase a ZOI.
> >
> > * SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI
> > delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
> >
> > * This would mean that the only case where WORD + SA + same
> > category as WORD + SI doesn't == nothing would be with ZO.
>
> * Furthermore, ZOI+SI == nothing, rather than being the beginning of
> a ZOI clause that requires a 5th dan Lojban black belt to get out
> of ("zoi .y. si .y. .y. si si si si si").

I too like this option. SA and SU should be disallowed as ZOI
delimiters as well.

BTW, is there any way to get out of a ZOI clause if you forget
what you used as an opening delimiter? You just start reciting the
dictionary?

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 06, 2004 at 07:17:22PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > > Option 3:
> > >
> > > * SI is NOT an allowed ZOI delimiter.
> > >
> > > * SI after ZOI eats through delimiters and text seperately, so
> > > four SI in a row are required to erase a ZOI.
> > >
> > > * SA+ZOI puts you back in the ZOI. Since SI can NOT be a
> > > ZOI delimiter, SA+ZOI+SI is the fastest way to delete the
> > > whole thing.
> > >
> > > * This would mean that the only case where WORD + SA + same
> > > category as WORD + SI doesn't == nothing would be with ZO.
> >
> > * Furthermore, ZOI+SI == nothing, rather than being the
> > beginning of a ZOI clause that requires a 5th dan Lojban black
> > belt to get out of ("zoi .y. si .y. .y. si si si si si").
>
> I too like this option. SA and SU should be disallowed as ZOI
> delimiters as well.

Doing so adds very little functionality, but yeah, might as well.

> BTW, is there any way to get out of a ZOI clause if you forget
> what you used as an opening delimiter? You just start reciting the
> dictionary?

/me chortles.

Yeah, I think you're screwed at that point.

-Robin


posts: 1912


> > I too like this option. SA and SU should be disallowed as ZOI
> > delimiters as well.
>
> Doing so adds very little functionality, but yeah, might as well.

Ideally SI, SA and SU will have the same precedences in all cases.
That way it's easier to learn the rules.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


>
> So you're saying that {broda ba'e bu} is a very lerfu broda?

Not exactly. When you say {broda ba'e bu} you are emphasizing
{bu}. For example, to make sure that the listener pays attention
and doesn't think you are using broda as a brivla. It's {broda *bu*}.
For example:

mi pu cusku lu broda ba'e bu li'u

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912



> Re: Magic Words
> grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU clauses. I'm
> pretty sure this is just a YACC side effect; it certainly seems amazingly
> ridiculous to me. If anyone disagrees, let me know.

I disagree.

> I suppose I can imagine something like "mi zoi zoi quietude zoi zei cinmo",
> but I don't know if it's worthwhile building such a bizarre construction into
> the language. Maybe.
>
> ZO-clause+ZEI and LOhU...LEhU+ZEI are just dumb as hell, though.

But allowing them gives a simpler grammar than disallowing them,
doesn't it? You don't need a special rule saying that ZEI is not
allowed after LEhU or after a word preceded by ZO. It's useful
for something like {zo zei zei lujvo}

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 06, 2004 at 07:52:06PM -0800, Jorge Llamb?as wrote:
> --- wikidiscuss@lojban.org wrote:
> > Re: Magic Words
> > grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU
> > clauses. I'm pretty sure this is just a YACC side effect; it
> > certainly seems amazingly ridiculous to me. If anyone disagrees,
> > let me know.
>
> I disagree.

OK.

> > I suppose I can imagine something like "mi zoi zoi quietude zoi
> > zei cinmo", but I don't know if it's worthwhile building such a
> > bizarre construction into the language. Maybe.
> >
> > ZO-clause+ZEI and LOhU...LEhU+ZEI are just dumb as hell, though.
>
> But allowing them gives a simpler grammar than disallowing them,
> doesn't it?

Not in PEG it doesn't. Not by a long shot. ZEI takes single words;
none of these things are single words.

> You don't need a special rule saying that ZEI is not allowed after
> LEhU or after a word preceded by ZO.

I don't need one now, either. In {zo broda zei brode}, both zo and
zei are trying to get broda. Error. No problem.

> It's useful for something like {zo zei zei lujvo}

Seems pretty marginal to me, as that's one of the only words that
zei can't already attach to. I can allow {zei zei lujvo} easily
enough regardless.

-Robin


posts: 14214

On Sat, Nov 06, 2004 at 08:36:38PM -0800, Robin Lee Powell wrote:
> On Sat, Nov 06, 2004 at 07:52:06PM -0800, Jorge Llamb?as wrote:
> > > I suppose I can imagine something like "mi zoi zoi quietude
> > > zoi zei cinmo", but I don't know if it's worthwhile building
> > > such a bizarre construction into the language. Maybe.
> > >
> > > ZO-clause+ZEI and LOhU...LEhU+ZEI are just dumb as hell,
> > > though.
> >
> > But allowing them gives a simpler grammar than disallowing them,
> > doesn't it?
>
> Not in PEG it doesn't. Not by a long shot. ZEI takes single
> words; none of these things are single words.

To be more specific: the rule for ZEI in my PEG grammar is
(simplified slightly):

any-word (ZEI any-word)+ indicators*

any-word matches a single Lojban word. Duh.

Anything else will require an explicit exception.

Note that currently, as a side effect, my parser is just fine with
{ti zei zei broda} and such.

Not that my parser, or any other parser, should be relevant to this
discussion, but he started it.

-Robin


posts: 14214

On Sat, Nov 06, 2004 at 07:52:06PM -0800, Jorge Llamb?as wrote:
>
> --- wikidiscuss@lojban.org wrote:
>
> > Re: Magic Words
> > grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU
> > clauses. I'm pretty sure this is just a YACC side effect; it
> > certainly seems amazingly ridiculous to me. If anyone disagrees,
> > let me know.
>
> I disagree.

OK. Tell me what is special about these constructs that something
that binds together words:

zei ZEI lujvo glue
joins preceding and following words into a
lujvo

should be able to bind these things, which are clearly not single
words, as well?

And if it can bind these things, why not lu...li'u, or tu'e...tu'u,
or any other grouped construct?

-Robin


posts: 1912


> > > grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU
> > > clauses. I'm pretty sure this is just a YACC side effect; it
> > > certainly seems amazingly ridiculous to me. If anyone disagrees,
> > > let me know.
> >
> > I disagree.
>
> OK. Tell me what is special about these constructs that something
> that binds together words:
>
> zei ZEI lujvo glue
> joins preceding and following words into a
> lujvo
>
> should be able to bind these things, which are clearly not single
> words, as well?

I am not too picky about the meaning of "word". The way I see it,
{zei} grabs the first thing it finds to its right and the first thing
it finds to its left and sticks them together to form a tanru unit.

If you are picky about what "word" means there, then {da zei de zei di}
should give error as well.

> And if it can bind these things, why not lu...li'u, or tu'e...tu'u,
> or any other grouped construct?

Because {zei} has precedence over those, so it will just grab
{li'u} or {tu'u} by itself.

I am not particularly interested in the *effects* of allowing {zei}
to work on these constructs. I just want the simplest possible rule.
The way I see it now, {zo} and {zoi} act first, so {zei} has just
their output to work with. Maybe there is a simpler way of looking
at it?

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 06, 2004 at 09:00:37PM -0800, Jorge Llamb?as wrote:
> I am not particularly interested in the *effects* of allowing
> {zei} to work on these constructs. I just want the simplest
> possible rule. The way I see it now, {zo} and {zoi} act first, so
> {zei} has just their output to work with.

OK, that's where we're stuck.

"output"? These aren't functions, these are strings of text. "zo
broda" is two words. We define the former as quoting the latter,
but there's no "output".

> Maybe there is a simpler way of looking at it?

My meta-rules (which I didn't know I had):

1. Magic words lay claim to, or "grab", other words, with varying
effects.

2. Magic word grabbing proceeds by an order of precedence. The
effect of a magic word can be to prevent the operation of another
magic word (zo in particular) in which case the word that has been
so grabbed no longer is allowed to grab other words.

3. No two magic words may grab the same word, unless the
interaction between those two magic words has been clearly defined.
Any attempt to do so is an error.

Reading {zo broda zei brode}, for example, we have:

zo grabs "broda". No other grabbing word is allowed to have it.

zei attempts to grab broda, but broda has already been grabbed. An
error results.

For {broda si zei brode} we have a similar error, assuming that si
is of higher precedence than zei, which I am.

-Robin


posts: 1912


> To be more specific: the rule for ZEI in my PEG grammar is
> (simplified slightly):
>
> any-word (ZEI any-word)+ indicators*
>
> any-word matches a single Lojban word. Duh.

Any word at all? What does {zo zei cmavo} give?

> Anything else will require an explicit exception.

I don't have much of a feeling for how the PEG grammar works yet.

> Note that currently, as a side effect, my parser is just fine with
> {ti zei zei broda} and such.

Isn't that {ti-zei-zei broda}?

> Not that my parser, or any other parser, should be relevant to this
> discussion, but he started it.

If it's simpler for the parser to do it one way, then probably
there is a simple way for us to understand the rule that way.
I don't as yet see how the PEG does it, so I don't see how the
rule would be simpler that way.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 06, 2004 at 09:23:15PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > To be more specific: the rule for ZEI in my PEG grammar is
> > (simplified slightly):
> >
> > any-word (ZEI any-word)+ indicators*
> >
> > any-word matches a single Lojban word. Duh.
>
> Any word at all? What does {zo zei cmavo} give?

zo wins, due to being farther up the food chain:

text
sentence
|- zoClause
| |- ZO: zo
| |- anyWord: zei
|- BRIVLA: cmavo

> > Note that currently, as a side effect, my parser is just fine
> > with {ti zei zei broda} and such.
>
> Isn't that {ti-zei-zei broda}?

Nope.

text
sentence
|- KOhA: ti
|- tanruUnit2
|- anyWord: zei
|- ZEI: zei
|- anyWord: cmavo

But your question reminds me of why Cowan and I decided that ZEI shouldn't act
on ZEI: there's no obvious choice for whether to group left or group right.
Thanks.

-Robin


posts: 1912


> But your question reminds me of why Cowan and I decided that ZEI shouldn't
> act
> on ZEI: there's no obvious choice for whether to group left or group right.
> Thanks.

Why isn't order of appearance the obvious choice?

{zo zo} works by order of appearance.
{zoi zoi} works by order of appearance.
{si si} works by order of appearance.
{bu bu} works by order of appearance, in a way.
{lo'u lo'u} works by order of appearance.

How come {zei zei} is the exception?

mu'o mi'e xorxes



__
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com


posts: 14214

On Sat, Nov 06, 2004 at 10:17:43PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > But your question reminds me of why Cowan and I decided that ZEI
> > shouldn't act on ZEI: there's no obvious choice for whether to
> > group left or group right. Thanks.
>
> Why isn't order of appearance the obvious choice?
>
> {zo zo} works by order of appearance.
> {zoi zoi} works by order of appearance.
> {lo'u lo'u} works by order of appearance.

Those do, yes.

> {si si} works by order of appearance.

No, it's a string that deletes the two previous words.

> {bu bu} works by order of appearance, in a way.

No, it's an error, at least by the beginning of text.

> How come {zei zei} is the exception?

It's hardly the only one.

The problem is, if you want to say "I said a zei lujvo", we have:

{mi pu cusku lo zei zei lujvo}

or something.

The problem is that if we take it by order of appearence, as you
say, the first zei grabs the lo and the second zei, so we end up
with "I was a speaker type-of lo type of zei type of lujvo".

-Robin


posts: 1912


> > {zo zo} works by order of appearance.
> > {zoi zoi} works by order of appearance.
> > {lo'u lo'u} works by order of appearance.
>
> Those do, yes.
>
> > {si si} works by order of appearance.
>
> No, it's a string that deletes the two previous words.

The first {si} erases the previous word.
Then comes the second, and erases its previous word, from
what's left after the first {si} has acted.

This may not be how it is implemented in your parser, but it's a valid
description of what goes on. If they didn't act by order of appearance,
the second {si} might erase the first one.

> > {bu bu} works by order of appearance, in a way.
>
> No, it's an error, at least by the beginning of text.

That's beside the point. {zei zei} could be an error at the
beginning of text too, since the first zei comes along and
has nothing in front to grab. But they still act in order of
appearance. The first {bu} changes the previous word into
a lerfu, and then the second {bu} changes it into some other
lerfu. If they didn't work by order of appearance, the second
{bu} might turn the first into a lerfu and thus disable its
function as a bu.

> > How come {zei zei} is the exception?
>
> It's hardly the only one.

It would be the only one as far as I can see.

> The problem is, if you want to say "I said a zei lujvo", we have:
>
> {mi pu cusku lo zei zei lujvo}
>
> or something.
>
> The problem is that if we take it by order of appearence, as you
> say, the first zei grabs the lo and the second zei, so we end up
> with "I was a speaker type-of lo type of zei type of lujvo".

Right. So {zei zei lujvo} is not the right way to say "zei-lujvo".
That's why {zo zei zei lujvo} would be useful. But even if
{zo zei zei} were not allowed, I see no reason to complicate ZEI by
not letting it work by order of appearance, like everything else. It
looks like an exception to me.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 06, 2004 at 10:39:15PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > > {zo zo} works by order of appearance.
> > > {zoi zoi} works by order of appearance.
> > > {lo'u lo'u} works by order of appearance.
> >
> > Those do, yes.
> >
> > > {si si} works by order of appearance.
> >
> > No, it's a string that deletes the two previous words.
>
> The first {si} erases the previous word. Then comes the second,
> and erases its previous word, from what's left after the first
> {si} has acted.

Ah, OK. Fair.

> > > {bu bu} works by order of appearance, in a way.
> >
> > No, it's an error, at least by the beginning of text.
>
> That's beside the point. {zei zei} could be an error at the
> beginning of text too, since the first zei comes along and has
> nothing in front to grab. But they still act in order of
> appearance. The first {bu} changes the previous word into a lerfu,
> and then the second {bu} changes it into some other lerfu. If they
> didn't work by order of appearance, the second {bu} might turn the
> first into a lerfu and thus disable its function as a bu.

OK.

> > > How come {zei zei} is the exception?
> >
> > It's hardly the only one.
>
> It would be the only one as far as I can see.

Well, SA SA is another, but that's maybe a bit different.

> > The problem is, if you want to say "I said a zei lujvo", we
> > have:
> >
> > {mi pu cusku lo zei zei lujvo}
> >
> > or something.
> >
> > The problem is that if we take it by order of appearence, as you
> > say, the first zei grabs the lo and the second zei, so we end up
> > with "I was a speaker type-of lo type of zei type of lujvo".
>
> Right. So {zei zei lujvo} is not the right way to say "zei-lujvo".

Then what the hell is it useful for? Seriously?

> That's why {zo zei zei lujvo} would be useful.

I just wrote that it.

> But even if {zo zei zei} were not allowed, I see no reason to
> complicate ZEI by not letting it work by order of appearance, like
> everything else. It looks like an exception to me.

True.

I hate to bring this up, but right-grouping (which is what you're
talking about) is very hard in PEGs. I'm sure it's possible, but
I've had some troubles.

Bizarrely enough, some examples of valid Lojban reverse polish
notation stuff, the more complicated the better, would help with
this quite a lot.

-Robin


posts: 14214

> > > > How come {zei zei} is the exception?
> > >
> > > It's hardly the only one.
> >
> > It would be the only one as far as I can see.
>
> Well, SA SA is another, but that's maybe a bit different.

Oh, it's not an exception, actually; I forgot about the multiple
consecutive SA feature. Adding that in now.

-Robin


posts: 14214

On Sat, Nov 06, 2004 at 10:53:31PM -0800, Robin Lee Powell wrote:
> On Sat, Nov 06, 2004 at 10:39:15PM -0800, Jorge Llamb?as wrote:
> > > The problem is, if you want to say "I said a zei lujvo", we
> > > have:
> > >
> > > {mi pu cusku lo zei zei lujvo}
> > >
> > > or something.
> > >
> > > The problem is that if we take it by order of appearence, as
> > > you say, the first zei grabs the lo and the second zei, so we
> > > end up with "I was a speaker type-of lo type of zei type of
> > > lujvo".
> >
> > Right. So {zei zei lujvo} is not the right way to say
> > "zei-lujvo".
>
> Then what the hell is it useful for? Seriously?

I have a compromise idea. Unless you can suggest something that
handling ZEI in this way would be useful for, why don't we just say
that any string of consecutive ZEI acts as though there were only
one? Call it emphasis.

> > That's why {zo zei zei lujvo} would be useful.
>
> I just wrote that it.

s/it/in/

> > But even if {zo zei zei} were not allowed, I see no reason to
> > complicate ZEI by not letting it work by order of appearance, like
> > everything else. It looks like an exception to me.
>
> True.
>
> I hate to bring this up, but right-grouping (which is what you're
> talking about) is very hard in PEGs. I'm sure it's possible, but
> I've had some troubles.

Errrm. Left grouping. Duh. The current version right groups. The

  • only* place that I've found in the grammar where bare left grouping

is actually required is the RPN stuff, and as I mentioned I haven't
tested that well.

-Robin


posts: 14214

On Sat, Nov 06, 2004 at 11:27:12PM -0800, Robin Lee Powell wrote:
> On Sat, Nov 06, 2004 at 10:53:31PM -0800, Robin Lee Powell wrote:
> > On Sat, Nov 06, 2004 at 10:39:15PM -0800, Jorge Llamb?as wrote:
> > > But even if {zo zei zei} were not allowed, I see no reason to
> > > complicate ZEI by not letting it work by order of appearance,
> > > like everything else. It looks like an exception to me.
> >
> > True.
> >
> > I hate to bring this up, but right-grouping (which is what
> > you're talking about) is very hard in PEGs. I'm sure it's
> > possible, but I've had some troubles.
>
> Errrm. Left grouping. Duh. The current version right groups.
> The *only* place that I've found in the grammar where bare left
> grouping is actually required is the RPN stuff, and as I mentioned
> I haven't tested that well.

I just re-tested it, and am in fact certain that it does *not* work
properly. It may only parse valid left-grouped RPN expressions, but
they certainly aren't parsed in a left-grouped fashion.

-Robin


posts: 14214

Deep in to talking to myself now.

On Sat, Nov 06, 2004 at 11:32:06PM -0800, Robin Lee Powell wrote:
> On Sat, Nov 06, 2004 at 11:27:12PM -0800, Robin Lee Powell wrote:
> > On Sat, Nov 06, 2004 at 10:53:31PM -0800, Robin Lee Powell
> > wrote:
> > > On Sat, Nov 06, 2004 at 10:39:15PM -0800, Jorge Llamb?as
> > > wrote:
> > > > But even if {zo zei zei} were not allowed, I see no reason
> > > > to complicate ZEI by not letting it work by order of
> > > > appearance, like everything else. It looks like an exception
> > > > to me.
> > >
> > > True.
> > >
> > > I hate to bring this up, but right-grouping (which is what
> > > you're talking about) is very hard in PEGs. I'm sure it's
> > > possible, but I've had some troubles.
> >
> > Errrm. Left grouping. Duh. The current version right groups.
> > The *only* place that I've found in the grammar where bare left
> > grouping is actually required is the RPN stuff, and as I
> > mentioned I haven't tested that well.
>
> I just re-tested it, and am in fact certain that it does *not*
> work properly.

I'm no longer certain this is true. In fact, I have yet to find
something that doesn't parse properly, but I'm really bad with RPN.
Having someone who isn't beat on it would be lovely.

> It may only parse valid left-grouped RPN expressions, but they
> certainly aren't parsed in a left-grouped fashion.

This is definately true, but maybe unavoidable. Certainly I'm not
good enough to fix it.

Regardless, the left-grouping ZEI change is a very different thing.
It was just a matter of removing a special case that told ZEI not to
work when there was a ZEI following.

I still think there's no point in allowing left-grouping or
right-grouping ZEI strings, and that if we allow them at all they
should be treated as equivalent to a single ZEI.

-Robin


On Saturday 06 November 2004 21:27, wikidiscuss@lojban.org wrote:
> Re: Magic Words
> The red book claims that UI cmavo can't mark BU. What's up with that?
>
> Specifically, it says:
>
> UI and CAI cmavo mark the previous word, except for ``zo, ``si, ``sa'',
> ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE cmavo, and ``bu''. Multiple
> UI cmavo may be used in succession. A following ``nai'' is made part of the
> UI.

If UI can mark BY, then it should mark the BY resulting from BU grabbing a
word. Thus in {.abu .ui viska le blabi ractu}, {ui} marks {abu}.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 1912


> Regardless, the left-grouping ZEI change is a very different thing.
> It was just a matter of removing a special case that told ZEI not to
> work when there was a ZEI following.

"Removing a special case" sounds very nice.

> I still think there's no point in allowing left-grouping or
> right-grouping ZEI strings, and that if we allow them at all they
> should be treated as equivalent to a single ZEI.

I see disallowing the natural formation as complicating the grammar
unnecessarily. I'm not arguing that one rule would give more useful
results than another, mostly this stuff will go unused anyway. All
I'm saying is that the simplest rule would seem to be that the first
{zei} removes the magic from the second, just because it comes first.
That's what magic words always do. Exceptions to that rule should
have a very good justification.

I presume {da zei zo} {da zei zoi} and {da zei lo'u} are all
allowed, even though they wouldn't work as lujvo the other way
around? {da zei zei} is exactly the same case, the first {zei}
removes any functionality from the following word.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




On Saturday 06 November 2004 21:49, wikidiscuss@lojban.org wrote:
> I have a strong preference for option three; it gives the most user
> friendliness, IMO. What do you all think?

I think SI, LOhU, and LEhU should all be disallowed as ZOI delimiters.
Disallowing a word, or a few words, as a delimiter is no hardship, as there
are infinitely many words, so for any finite quoted string, there is a word
not contained in it.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


On Saturday 06 November 2004 22:17, Robin Lee Powell wrote:
> On Sat, Nov 06, 2004 at 10:14:38PM -0500,
>
> MorphemeAddict@wmconnect.com wrote:
> > In a message dated 2004-11-06 9:29:38 PM Eastern Standard Time,
> >
> > robin via wikidiscuss@lojban.org writes:
> > > UI and CAI cmavo mark the previous word, except for ``zo'',
> > > ``si, ``sa, ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE
> > > cmavo, and ``bu''.
> >
> > Looks like "uibu" is okay, but not "bu.ui".
>
> Well, yes. My question was, why?

{bu.ui} at the beginning of a sentence is invalid because {bu} has nothing to
grab. {.abu.ui} is valid, according to jbofi'e.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


> Option 3:
>
> * SI is NOT an allowed ZOI delimiter.
> * SI after ZOI eats through delimiters and text seperately, so four SI in a row are required to erase a ZOI.
> * SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
> * This would mean that the only case where WORD + SA + same category as WORD + SI doesn't == nothing would be with ZO.
>
> I have a strong preference for option three; it gives the most user friendliness, IMO. What do you all think?

I very much like option 3. I'm curious about one point you mentioned earlier,
and whether it's still a problem, though. After erasing the final delimiter,
are we in a place to add more quoted text? ISTR you saying that

zoi zoi I'm talking zoi si about quotes zoi

was invalid, or at least different from

zoi zoi I'm talking about quotes zoi

because although the si ate the terminating zoi, the first zoi had already
turned it into something else and stopped working. Am I misremember, is that
somehow fixed, or what? (I really hope it is, otherwise requiring multiple
{si}s would be really weird).

Thanks for trying to comprehend all these interactions. I'm just barely able
to keep my head from exploding on many of them.
--
Adam Lopresto
http://cec.wustl.edu/~adam/

I am a peripheral visionary; I can see into the future, but only way off to the sides.


posts: 14214

This isn't bad, except for:

On Sun, Nov 07, 2004 at 11:29:32AM -0800, wikidiscuss@lojban.org wrote:
> Definition: Y is not a word.

Not a good idea. Inherently contradictory, for one thing.

> ZEI: Unless it has been turned off by a preceding magic word, it
> takes the previous word or magic word output,
>
snip
>
> BU: Unless it has been turned off by a preceding magic word, it
> takes the previous word or magic word output,
>
snip
>
> SI: Unless it has been turned off by a preceding magic word, it
> takes the previous word or magic word output
>
snip

Again, where does this "magic word output" concept come from? These
aren't functions, they are streams of text to which we are applying
semantic meaning. Why not just stick with words?

-Robin


posts: 14214

On Sun, Nov 07, 2004 at 11:29:32AM -0800, wikidiscuss@lojban.org wrote:
> Re: Magic Words
> The exeptionless rules for magic words would be as follows:

You forgot UI and friends.

-Robin


posts: 14214

> Again, where does this "magic word output" concept come from?
> These aren't functions, they are streams of text to which we are
> applying semantic meaning. Why not just stick with words?

To give a concrete example of why this bothers me, the exceptionless
rules cause an error for:

broda broda si bu

because BU acts on the "output" of the previous magic word, si,
which is nothing, but BU isn't allowed to act on nothing, so an
error results.

The other problem is that writing up definitions based on the
exceptionless rules pretty much requires enshrining the term "magic
word".

I am not completely averse to the exceptionless rules (they are
quite simple, after all), but I'd really prefer to stick with real
things in the text stream (i.e. words) if we can.

-Robin


posts: 1912


> This isn't bad, except for:
>
> > Definition: Y is not a word.
>
> Not a good idea. Inherently contradictory, for one thing.

By that I mean it is invisible to magic words. {Y BU} would constitute an
exception.

> > ZEI: Unless it has been turned off by a preceding magic word, it
> > takes the previous word or magic word output,
> >
> snip
> >
> > BU: Unless it has been turned off by a preceding magic word, it
> > takes the previous word or magic word output,
> >
> snip
> >
> > SI: Unless it has been turned off by a preceding magic word, it
> > takes the previous word or magic word output
> >
> snip
>
> Again, where does this "magic word output" concept come from? These
> aren't functions, they are streams of text to which we are applying
> semantic meaning. Why not just stick with words?

Because sticking with words creates conflicts when more than one
magic word vie for the same word. Taking the whole magic word construct
as an "honorary word" for other magic words solves all those conflicts.
If {bu} can act on a preceding {a bu} and turn that into a lerfu, there
is no reason why it can't similarly not act on {zo a} or on {a zei e} or
on {lo'u a le'u}. The reason these are special constructs is
just that they involve magic words. Instead of saying that {bu} fails
because other magic words have already staked a claim the word it should
change into a lerfu, we say that what it changes into a lerfu is the
previous word or magic-word-construct.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> To give a concrete example of why this bothers me, the exceptionless
> rules cause an error for:
>
> broda broda si bu
>
> because BU acts on the "output" of the previous magic word, si,
> which is nothing, but BU isn't allowed to act on nothing, so an
> error results.

No. The "output" of si is not nothing. SI erases the previous word
and leaves the one before that as the last word of the speech stream.
So its output is the word before last.

> The other problem is that writing up definitions based on the
> exceptionless rules pretty much requires enshrining the term "magic
> word".

I'm not sure this would be necessary. But it does require the notion
of "honorary word", or "pseudo-word" or some such. In the case of
{broda broda si bu}, {si} leaves the honorary word {broda brode si}
which has the same meaning as {broda}, for {bu} to act on.

> I am not completely averse to the exceptionless rules (they are
> quite simple, after all), but I'd really prefer to stick with real
> things in the text stream (i.e. words) if we can.

But honorary words are just as real as words! {lo'u} can take more
than one word in its scope, why is it such a big deal that other
magic words take more than one word in some cases as well?

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

> > I am not completely averse to the exceptionless rules (they are
> > quite simple, after all), but I'd really prefer to stick with
> > real things in the text stream (i.e. words) if we can.
>
> But honorary words are just as real as words! {lo'u} can take more
> than one word in its scope, why is it such a big deal that other
> magic words take more than one word in some cases as well?

Two reasons:

1. It's a substantial change to the language.

2. I have no idea what most of these constructs would actually

  • mean*.


-Robin


posts: 14214

On Sun, Nov 07, 2004 at 01:28:36PM -0600, Adam D. Lopresto wrote:
> >Option 3:
> >
> >* SI is NOT an allowed ZOI delimiter.
> >* SI after ZOI eats through delimiters and text seperately, so four SI in
> >a row are required to erase a ZOI.
> >* SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI
> >delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
> >* This would mean that the only case where WORD + SA + same category as
> >WORD + SI doesn't == nothing would be with ZO.
> >
> >I have a strong preference for option three; it gives the most
> >user friendliness, IMO. What do you all think?
>
> I very much like option 3. I'm curious about one point you
> mentioned earlier, and whether it's still a problem, though.
> After erasing the final delimiter, are we in a place to add more
> quoted text?

Yes, absolutely.

> ISTR you saying that
>
> zoi zoi I'm talking zoi si about quotes zoi
>
> was invalid, or at least different from
>
> zoi zoi I'm talking about quotes zoi
>
> because although the si ate the terminating zoi, the first zoi had
> already turned it into something else and stopped working.

That was when I was using a pre-processor. In the default mode, my
current parser outputs the following for the first one:

text
zoiClause
|- ZOI: zoi
|- anyWordZoi: zoi
|- zoiPrintableWord: I'm
|- zoiWords
|- zoiPrintableWord: talking
|- zoiWords
|- zoiPrintableWord: about
|- zoiWords
|- zoiPrintableWord: quotes
|- anyWordZoi: zoi

> Thanks for trying to comprehend all these interactions. I'm just
> barely able to keep my head from exploding on many of them.

Thanks for the encouragement. :-)

xorxes is working hard on this too. Thanks, xorxes.

-Robin


posts: 14214

On Sun, Nov 07, 2004 at 10:34:25AM -0500, Pierre Abbat wrote:
> On Saturday 06 November 2004 22:17, Robin Lee Powell wrote:
> > On Sat, Nov 06, 2004 at 10:14:38PM -0500,
> >
> > MorphemeAddict@wmconnect.com wrote:
> > > In a message dated 2004-11-06 9:29:38 PM Eastern Standard
> > > Time,
> > >
> > > robin via wikidiscuss@lojban.org writes:
> > > > UI and CAI cmavo mark the previous word, except for ``zo'',
> > > > ``si, ``sa, ``su, ``lo'u, ZOI, ``fa'o, ``zei,
> > > > BAhE cmavo, and ``bu''.
> > >
> > > Looks like "uibu" is okay, but not "bu.ui".
> >
> > Well, yes. My question was, why?
>
> {bu.ui} at the beginning of a sentence is invalid because {bu} has
> nothing to grab. {.abu.ui} is valid, according to jbofi'e.

Well, yes, but the Red Book says that .ui can't modify bu *in*

  • general*. This implies to me that "broda bu .ui" is invalid.


-Robin


On Sunday 07 November 2004 14:29, wikidiscuss@lojban.org wrote:
> SA: Unless it has been turned off by a preceding magic word, it deletes
> everything back up to and including the first word found of the selmaho of
> the word that follows (stuff processed by preceding magic words in general
> does not keep its original selmaho) or else if no such word is found, up to
> the beginning of text. If no word follows, it gives an error.
>
> BAhE: Unless it has been turned off by a preceding magic word, it modifies
> the word that follows, and becomes invisible (i.e. it is not part of the
> previous stuff for following magic words). If what follows is a magic word,
> it does not turn it off. If no word follows, it gives an error.

ti ba'e skargolu la karsarli sa za'e mantanko la marcuigli

({mantanko} here means "mundungus", in the sense of a kind of tobacco.)

mu'omi'e pier.
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 1912



> > > I am not completely averse to the exceptionless rules (they are
> > > quite simple, after all), but I'd really prefer to stick with
> > > real things in the text stream (i.e. words) if we can.
> >
> > But honorary words are just as real as words! {lo'u} can take more
> > than one word in its scope, why is it such a big deal that other
> > magic words take more than one word in some cases as well?
>
> Two reasons:
>
> 1. It's a substantial change to the language.

Not really. In many cases, that's how it works now: {a bu bu} already
follows that concept.

> 2. I have no idea what most of these constructs would actually
> *mean*.

That's the easy part. {<anything> bu} is absolutely conventional.
Giving a meaning to {zo a bu} is no harder than giving a meaning
to {cei bu}. And we don't *have* to give it any meaning other
than say it is a lerfu. The same applies to {zei}. Why would
{zo a zei da} present more difficulties in meaning than
{cei zei da}?

The only magic words that would work with magic word constructs
are BU, ZEI, SI and to some extent SA. The rest don't look back,
so constructs don't affect them as input.

(I'm not sure what to say about UIs. It doesn't look like they
are much like magic words.)

Let's see how much the exceptionless rules differ from the
prescription:

> Y is completely ignored (i.e. considered whitespace) except before BU.

Same thing. i.e. {Y BU} would have to be an exception.

> ZO quotes the following word, no matter what it is, except Y. Words
> quoted with ZO lose their grammatical functions.

No change.

> FAhO terminates the word or text stream unequivocably, unless
> quoted with ZO.

The exceptionless rules also allow {da zei fa'o}, {zoi fa'o ... fa'o},
and {lo'u ... fa'o ... le'u}, in which the magic of {fa'o} has been
turned off by a preceding magic word.

> LOhU quotes all following Lojban words up to a LEhU (but not a ZO+LEhU;
> this is to allow nested LOhU...LEhU quotes inside a LOhU...LEhU, so you
> can talk about mistakes you made that include a previous error quote).
> Except for the ZO+LEhU case (which should be read as simply LEhU inside
> a LOhU...LEhU quote) all Lojban words within a LOhU...LEhU quote are
> read without any grammar and have no grammatical effects.

The exceptionles rules make no exception for {zo le'u}. You don't
mention the exception made for {fa'o} here.

> LEhU is ungrammatical except at the end of a LOhU quotation and
> after ZO.

The exceptionless rules will also allow {zoi le'u ... le'u} and
{da zei le'u}, in which the magic of LEhU is turned off by a
preceding magic word.

> SI erases the preceding word unless it is a ZO. Y is ignored.

This is probably the most different for the exceptionless rules.

SI would NOT erase the previous word in the following cases:
{zo si}, {zoi si ... si}, {da zei si}, {lo'u ... si ... le'u},
where the magic of SI is turned off by a preceding magic word.

SI would erase MORE than a single word in the following cases:
{zo a si}, {zoi zoi ... zoi si}, {da zei de si},
{lo'u ... le'u si}, {a bu si}, {... sa si}, where SI erases
the whole preceding magic word construct.

> SA erases the preceding word and other words, unless the preceding
> word is a ZO. SA erases back until it sees a word of the same
> selma'o as the word that follows SA. The previous same-selma'o word
> is itself erased. Y is ignored for selma'o matching purposes.

With the exceptionless rules, {sa} doesn't erase anything in
{zo sa}, {zoi sa ... sa}, {lo'u ... sa ... le'u}, {da zei sa},
where its magic has been turned off by a preceding magic word.

How is the selmaho rule supposed to work with respect to words
grabbed by magic words? Do they count for SA? For example:

mi cusku zo broda sa bacru

Does {bacru} replace {cusku} or {broda}? (With the exceptionless
rules, it would replace {cusku}, because {broda} is hidden within
a magic word construct, so SA doesn't see it.)

> SU erases itself and all words back to the beginning of the
> current conversation for the speaker that says it. If spoken twice,
> it also erases the other speaker's words. In computer input cases
> SU, in general, erases itself and all words to the beginning of the
> input (as most computer input cases cannot distinguish between speakers).

With exceptionless rules, SU can appear deactivated by a preceding
magic word in: {zo su}, {zoi su ... su}, {lo'u ... su ...le'u} and
{da zei su}.

> ZOI cmavo use the following word as a delimiting word, no matter
> what it is, execept Y (which is ignored); and SI, SA and SU (which
> erase it).

With exceptionless rules there would be no exceptions. (Y not counting
as a word.)

ZOI is deactivated in {zo zoi}, {zoi zoi ... zoi}, {lo'u ... zoi ... le'u}
and {da zei zoi}.

> ZEI combines the preceding and the following word into a lujvo. For
> words to its left, it does not affect SI, SA, and SU (it affects
> whatever is to the left after the erasing is done; if nothing is left
> an error result);

This appears to be the same with exceptionless rules:

{da de si zei di} is a tanru unit. {zei} attaches the preceding
construct {da de si} to the following word {di} to form a tanru
unit. Of course, the meaning of the construct {da de si} is the
same as that of the word {da}.

> ZO and LOhU (which quote it); ZOI (which uses it
> as a delimiter);

That would be the same: {zo zei}, {lo'u ... zei ... le'u},
{zoi zei ... zei}. But {zo}, {zoi} and {lo'u} are allowed
as the second part of a zei: {da zei zo}, {da zei lo'u},
{da zei zoi}.

> BAhE and Y (which it skips);

In {da ba'e zei de}, {ba'e} modifies {zei} but does
not turn off its magic, so that's the same. But {ba'e} is
allowed as the second part of zei: {da zei ba'e}, where
{ba'e}'s magic is turned off by {zei}.

> ZEI (which would lead
> to grouping issues, and hence is an error to attempt);

No grouping issues with the exceptionless rules: {da zei zei}
is perfectly acceptable because the first zei turns off the
second.

> and FAhO (which
> makes no sense because the stream ends at the FAhO).

{da zei fa'o} is acceptable with the exceptionless rules because
zei turns off the magic of fa'o.

> For words to its
> right, it does not affect SI, SA, and SU (which erase it); BAhE and Y
> (which it skips); ZEI (which would lead to grouping issues, and hence
> is an error to attempt); and FAhO (which results in an error).

With exceptionless rules, SI, SA, SU, BAhE, ZEI and FAhO are instead
turned off by the preceding ZEI.

> ZO,
> ZOI and LOhU...LEhU clauses are attached to the ZEI lujvo in their
> entirety.

With exceptionless rules, that's the case if they are to the left of
ZEI. To the right, they are treated as a single word because their
magic is deactivated before they have a chance to make the clause.

> BAhE effects ZEI as usual, so ZEI skips over any preceeding
> BAhE cmavo to affect the word before them.

To the left, yes. To the right, BAhE is deactivated by ZEI.

mu'o mi'e xorxes





BU makes the preceding word into a lerfu word, except for ZO, ZEI and LOhU
(which quote it), ZOI (which uses it as a delimiter); LEhU (which would result
in bizarrely re-opening the LOhU...LEhU, and hence is an error); SI, SA and SU
(it affects whatever is to the left after the erasing is done; if nothing is
left an error results); BAhE (which it skips); BU (which would lead to grouping
issues, and hence is an error to attempt); and FAhO (after which anything,
including BU, is ignored). Note that Y is specifically included. BAhE effects
BU as usual, so BU skips over any preceeding BAhE cmavo to affect the word
before them. Multiple BU may be used in succession, in which case a new
letteral is formed for each additional BU (i.e. "broda bu" is a different
letteral from "broda bu bu"). However, "bu bu" by itself is illegal.







__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


I forgot BU in making the comparison of the exceptionless rules to the
prescription:

> BU makes the preceding word into a lerfu word, except for ZO, ZEI and LOhU
> (which quote it), ZOI (which uses it as a delimiter);

That's the same {zo bu}, {da zei bu}, {lo'u .. bu .. le'u} and
{zoi bu ... bu} all deactivate {bu}.

> LEhU (which would
> result
> in bizarrely re-opening the LOhU...LEhU, and hence is an error);

{lo'u ... le'u bu} is allowed by the exceptionless rules, but the
whole construct is turned into a lerfu. {le'u bu} by itself without
an opening {lo'u} is an error, but the error occurs before {bu}
appears, so it is caused by {le'u}, not by {bu}.

> SI, SA and SU
> (it affects whatever is to the left after the erasing is done;
> if nothing is left an error results);

That's the same.

This can also be described by saying that in {da de si bu}, {bu} acts
on the construct {da de si} and not just on the immediately preceding
word {si}.

> BAhE (which it skips); BU (which would lead to grouping
> issues, and hence is an error to attempt); and FAhO (after which anything,
> including BU, is ignored).

Those would work in the same way.

> Note that Y is specifically included.

That would have to be added as an exception.

> BAhE effects
> BU as usual, so BU skips over any preceeding BAhE cmavo to affect the word
> before them. Multiple BU may be used in succession, in which case a new
> letteral is formed for each additional BU (i.e. "broda bu" is a different
> letteral from "broda bu bu"). However, "bu bu" by itself is illegal.

Same.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 381

Since "si" erases the preceding word, what does it erase in the following
sentence:

Mi prami pendo .y si do

Does it erase ".y" or "pendo" or something else?

stevo

posts: 84

Someone else might have answered this; I'm wading through it...

wikidiscuss@lojban.org (that is, Robin) wrote:

>Re: Magic Words
>The red book claims that UI cmavo can't mark BU. What's up with that?
>
>Specifically, it says:
>
>UI and CAI cmavo mark the previous word, except for ``zo, ``si, ``sa, ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE cmavo, and ``bu. Multiple UI cmavo may be used in succession. A following ``nai'' is made part of the UI.
>
>
>
That can't be right. Then you couldn't say {mi viska .abu .ui} for
saying that I see my beloved Mr. A. {bu} is only magical to the left,
on its right side it's normal. I'm in favor of ruling this a
typographical error.

~mark


posts: 84

Robin Lee Powell wrote:

>On Sat, Nov 06, 2004 at 06:49:00PM -0800, wikidiscuss@lojban.org wrote:
>
>
>>Re: Magic Words
>>What do we do about SI, SA and ZOI?
>>
>>
>
>Forgot some issues here:
>
>
>
>>Option 3:
>>
>>* SI is NOT an allowed ZOI delimiter.
>>
>>* SI after ZOI eats through delimiters and text seperately, so
>>four SI in a row are required to erase a ZOI.
>>
>>* SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI
>>delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
>>
>>* This would mean that the only case where WORD + SA + same
>>category as WORD + SI doesn't == nothing would be with ZO.
>>
>>
>
>* Furthermore, ZOI+SI == nothing, rather than being the beginning of
>a ZOI clause that requires a 5th dan Lojban black belt to get out
>of ("zoi .y. si .y. .y. si si si si si").
>
>
OK. I was preferring option 2, since it seemed most consistent and
helpful (why do we need to eat text and delimiters separately anyway?)
but your point is a good one. Still, it's a bit of a wart, adding
another "any-word" that's not QUITE any word (though .y. is already
excluded).

~mark


posts: 84

Robin Lee Powell wrote:

>On Sat, Nov 06, 2004 at 07:17:22PM -0800, Jorge Llamb?as wrote:
>
>
>>BTW, is there any way to get out of a ZOI clause if you forget
>>what you used as an opening delimiter? You just start reciting the
>>dictionary?
>>
>>
>
>/me chortles.
>
>Yeah, I think you're screwed at that point.
>
>
That is a fabulous question, and we have to ponder. On one hand, I
would like there to be some emergency escape hatch. On the other, that
would mean something that needs escaping inside ZOI.

I can't shake the feeling that we do need to worry about this though.

~mark



On Sunday 07 November 2004 20:22, MorphemeAddict@wmconnect.com wrote:
> Since "si" erases the preceding word, what does it erase in the following
> sentence:
>
> Mi prami pendo .y si do
>
> Does it erase ".y" or "pendo" or something else?

What it should do is erase "pendo", since "y" is just a hesitation. In
jbofi'e, though, it erases "y".

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 381

In a message dated 2004-11-07 9:23:40 PM Eastern Standard Time,
phma@phma.hn.org writes:


> What it should do is erase "pendo", since "y" is just a hesitation. In
> jbofi'e, though, it erases "y".
>
> phma
>

Hesitation is its function, but it is an articulated morpheme, not just
silence (a hesitation). So is it a word here or not?

(And I like my example better as "Mi prami .y. si pendo do") I would expect
it to take out both the ".y." and "prami".

Of course all of these erasure words seem to ignore the fact that whatever is
to be erased was in fact actually spoken, and so can NOT be erased from the
mind of the listener. It's like telling a jury to disregard someone's
testimony or statement. They can try, but they have still heard it, and it can still
color their judgment.

stevo

stevo

wikidiscuss@lojban.org scripsit:
> Re: Magic Words
> The red book claims that UI cmavo can't mark BU. What's up with that?

Because the bu has already been absorbed to produce the lerfu. So
abu .ui means I'm happy about the referent, not about the bu.

--
"You know, you haven't stopped talking John Cowan
since I came here. You must have been http://www.reutershealth.com
vaccinated with a phonograph needle." jcowan@reutershealth.com
--Rufus T. Firefly http://www.ccil.org/~cowan


posts: 14214

On Sun, Nov 07, 2004 at 04:22:08PM -0500, Pierre Abbat wrote:
> On Sunday 07 November 2004 14:29, wikidiscuss@lojban.org wrote:
> > SA: Unless it has been turned off by a preceding magic word, it
> > deletes everything back up to and including the first word found
> > of the selmaho of the word that follows (stuff processed by
> > preceding magic words in general does not keep its original
> > selmaho) or else if no such word is found, up to the beginning
> > of text. If no word follows, it gives an error.
> >
> > BAhE: Unless it has been turned off by a preceding magic word,
> > it modifies the word that follows, and becomes invisible (i.e.
> > it is not part of the previous stuff for following magic words).
> > If what follows is a magic word, it does not turn it off. If no
> > word follows, it gives an error.
>
> ti ba'e skargolu la karsarli sa za'e mantanko la marcuigli
>
> ({mantanko} here means "mundungus", in the sense of a kind of
> tobacco.)

Umm, and your point would be?

-Robin


posts: 14214

On Sun, Nov 07, 2004 at 09:34:42PM -0500, John Cowan wrote:
> wikidiscuss@lojban.org scripsit:
> > Re: Magic Words The red book claims that UI cmavo can't mark BU.
> > What's up with that?
>
> Because the bu has already been absorbed to produce the lerfu. So
> a bu .ui means I'm happy about the referent, not about the bu.

You mean happy about the lerfu formed by A+BU, but not the BU,
correct?

IOW, de-reference "referent", please.

-Robin


posts: 14214

On Sun, Nov 07, 2004 at 08:34:27PM -0500, Mark E. Shoulson wrote:
> Robin Lee Powell wrote:
>
> >On Sat, Nov 06, 2004 at 07:17:22PM -0800, Jorge Llamb?as wrote:
> >
> >
> >>BTW, is there any way to get out of a ZOI clause if you forget
> >>what you used as an opening delimiter? You just start reciting
> >>the dictionary?
>
snip
>
> I can't shake the feeling that we do need to worry about this
> though.

I, on the other hand, find it an amazingly pointless thing to worry
about. First of all, most speakers use the same thing for the
delimiter 99% of the time (zoi is good for this; doesn't come up a
lot). For another, imagine how hostile your listener would have to
be to *not* cope with:

mi pu vitke la'o zoi The Great Cathedral Of The Infinite Light Of
The Procession Of Heaven .y. .y. .u'u mi na morji ma kau poi valsi
gi'e fanmo zo zoi

I mean, seriously. Anyone who can't put up with that is a dickhead.
In the computer case, you should be able to look at previous text.

If a situation comes up where you *must* have an out (verbal
computer input, for example), define a non-Lojban word for that
context that always leaves a zoi ("antidisestablismentarianism", for
example).

-Robin


On Sunday 07 November 2004 22:11, Robin Lee Powell wrote:
> On Sun, Nov 07, 2004 at 04:22:08PM -0500, Pierre Abbat wrote:
> > ti ba'e skargolu la karsarli sa za'e mantanko la marcuigli
> >
> > ({mantanko} here means "mundungus", in the sense of a kind of
> > tobacco.)
>
> Umm, and your point would be?

{sa za'e} appears to be an error, rather than erasing back to {ba'e}, since
the rule makes it invisible.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


On Sun, Nov 07, 2004 at 07:17:59PM -0800, Robin Lee Powell wrote:
> I, on the other hand, find it an amazingly pointless thing to worry
> about.

Indeed.

> If a situation comes up where you *must* have an out (verbal
> computer input, for example), define a non-Lojban word for that
> context that always leaves a zoi ("antidisestablismentarianism", for
> example).

Or click it in the ass.

--
Jay Kominek <jkominek@miranda.org>
UNIX is all about covering up
the fact that you can't type.


posts: 14214

On Sun, Nov 07, 2004 at 10:20:30PM -0500, Pierre Abbat wrote:
> On Sunday 07 November 2004 22:11, Robin Lee Powell wrote:
> > On Sun, Nov 07, 2004 at 04:22:08PM -0500, Pierre Abbat wrote:
> > > ti ba'e skargolu la karsarli sa za'e mantanko la marcuigli
> > >
> > > ({mantanko} here means "mundungus", in the sense of a kind of
> > > tobacco.)
> >
> > Umm, and your point would be?
>
> {sa za'e} appears to be an error, rather than erasing back to
> {ba'e}, since the rule makes it invisible.

Yes, but the rule operates left-to-right.

This means, by extension, that you can't put SI, SA or SU in a BU,
but I see no problem there.

-Robin


On Sunday 07 November 2004 22:27, Robin Lee Powell wrote:
> On Sun, Nov 07, 2004 at 10:20:30PM -0500, Pierre Abbat wrote:
> > On Sunday 07 November 2004 22:11, Robin Lee Powell wrote:
> > > On Sun, Nov 07, 2004 at 04:22:08PM -0500, Pierre Abbat wrote:
> > > > ti ba'e skargolu la karsarli sa za'e mantanko la marcuigli
> > >
> > > Umm, and your point would be?
> >
> > {sa za'e} appears to be an error, rather than erasing back to
> > {ba'e}, since the rule makes it invisible.
>
> Yes, but the rule operates left-to-right.

Yes. {ba'e} is on the left, so it has already been made invisible by the rule
by the time the preprocessor gets to {sa}, so {sa} has no BAhE to erase back
to. If {za'e} were invisible, {sa} would erase back to the previous brivla,
which is {karsarli}.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 14214

On Sun, Nov 07, 2004 at 10:54:24PM -0500, Pierre Abbat wrote:
> On Sunday 07 November 2004 22:27, Robin Lee Powell wrote:
> > On Sun, Nov 07, 2004 at 10:20:30PM -0500, Pierre Abbat wrote:
> > > On Sunday 07 November 2004 22:11, Robin Lee Powell wrote:
> > > > On Sun, Nov 07, 2004 at 04:22:08PM -0500, Pierre Abbat
> > > > wrote:
> > > > > ti ba'e skargolu la karsarli sa za'e mantanko la marcuigli
> > > >
> > > > Umm, and your point would be?
> > >
> > > {sa za'e} appears to be an error, rather than erasing back to
> > > {ba'e}, since the rule makes it invisible.
> >
> > Yes, but the rule operates left-to-right.
>
> Yes. {ba'e} is on the left, so it has already been made invisible
> by the rule by the time the preprocessor gets to {sa}, so {sa} has
> no BAhE to erase back to. If {za'e} were invisible, {sa} would
> erase back to the previous brivla, which is {karsarli}.

  • OOOH*.


Misunderstood, sorry.

Yes, that seems to be a bit of a problem.

-Robin


posts: 14214

On Sun, Nov 07, 2004 at 07:57:40PM -0800, Robin Lee Powell wrote:
> On Sun, Nov 07, 2004 at 10:54:24PM -0500, Pierre Abbat wrote:
> > On Sunday 07 November 2004 22:27, Robin Lee Powell wrote:
> > > On Sun, Nov 07, 2004 at 10:20:30PM -0500, Pierre Abbat wrote:
> > > > On Sunday 07 November 2004 22:11, Robin Lee Powell wrote:
> > > > > On Sun, Nov 07, 2004 at 04:22:08PM -0500, Pierre Abbat
> > > > > wrote:
> > > > > > ti ba'e skargolu la karsarli sa za'e mantanko la
> > > > > > marcuigli
> > > > >
> > > > > Umm, and your point would be?
> > > >
> > > > {sa za'e} appears to be an error, rather than erasing back
> > > > to {ba'e}, since the rule makes it invisible.
> > >
> > > Yes, but the rule operates left-to-right.
> >
> > Yes. {ba'e} is on the left, so it has already been made
> > invisible by the rule by the time the preprocessor gets to {sa},
> > so {sa} has no BAhE to erase back to. If {za'e} were invisible,
> > {sa} would erase back to the previous brivla, which is
> > {karsarli}.
>
> *OOOH*.
>
> Misunderstood, sorry.
>
> Yes, that seems to be a bit of a problem.

In fact, ba'e will cause many problems, even with left-to-right
ordering, which is why I made a seperate section ("Marking Words")
for it and UI.

{broda ba'e si broda} does very much the wrong thing under the
Exceptionless Rules, for example; the result is {broda *si* broda},
so si stays in the speech stream.

-Robin


Robin Lee Powell scripsit:

> {broda ba'e si broda} does very much the wrong thing under the
> Exceptionless Rules, for example; the result is {broda *si* broda},
> so si stays in the speech stream.

Hmm. I read the rules as generating {broda} (the first broda, that is),
as while ba'e tags the following word as emphatic, it does not
prevent it from having whatever effect it has.

Which means you can't erase a BAhE with "si". Which sucks.

BTW, it should say somewhere that after erasure words have done their
work, any remaining erasure words make the utterance ungrammatical.

--
"May the hair on your toes never fall out!" John Cowan
--Thorin Oakenshield (to Bilbo) jcowan@reutershealth.com


posts: 1912


> For another, imagine how hostile your listener would have to
> be to *not* cope with:
>
> mi pu vitke la'o zoi The Great Cathedral Of The Infinite Light Of
> The Procession Of Heaven .y. .y. .u'u mi na morji ma kau poi valsi
> gi'e fanmo zo zoi

Any competent speaker should cope with that, which clearly parses
as:

mi pu vitke la'o zoi <foreign noise including some parts
that sound like lojban> zoi

:-)

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> In fact, ba'e will cause many problems, even with left-to-right
> ordering, which is why I made a seperate section ("Marking Words")
> for it and UI.

That's an important point. If we define "magic word" as any word
that deals with the construct "any-word", then UI and BAhE are
not magic words. They do deal with every word in the same way,
but they don't deal with the construct "any-word" or anything
like it, which disables the ordinary function of any word.

So if I'm not mistaken for the true magic words we
would have:

zo-word = ZO any-word
zei-word = any-word ZEI any-word
bu-word = any-word BU
lohu-word = lo'u lehu-close
lehu-close = LEhU | any-word lehu-close
si-null = any-word SI
su-null = SU | any-word su-null
zoi-word =zoi any-word anything that-word
(I'm not sure how this one is formalized)

SA has to be handled selmaho by selmaho.

Where:

any-word = zo-word | zei-word | bu-word | lo'u-word
| zoi-word | A | ... | ZOhU

Y is not included in "any-word".

BAhE becomes part of the word that follows (except when it has
been transformed to "any-word" by the above rules, but that is
nothing special about BAhE, LE also doesn't convert the following
selbri into a sumti when it has been turned into "any-word").

> {broda ba'e si broda} does very much the wrong thing under the
> Exceptionless Rules, for example; the result is {broda *si* broda},
> so si stays in the speech stream.

{broda ba'e si brode} would be an emphatically deleted {broda},
replaced by {brode}: i.e. {ba'e} is always (no exceptions)
absorbed by the following word. The construct {ba'e si} would
be an emphasized word eraser.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 84

Jorge "Llamb����������������������������������" wrote:

(Wonder why I get that weird encoding)

>So if I'm not mistaken for the true magic words we
>would have:
>
>zo-word = ZO any-word
>zei-word = any-word ZEI any-word
>bu-word = any-word BU
>lohu-word = lo'u lehu-close
>lehu-close = LEhU | any-word lehu-close
>si-null = any-word SI
>su-null = SU | any-word su-null
>zoi-word =zoi any-word anything that-word
> (I'm not sure how this one is formalized)
>
>SA has to be handled selmaho by selmaho.
>
>Where:
>
>any-word = zo-word | zei-word | bu-word | lo'u-word
> | zoi-word | A | ... | ZOhU
>
>Y is not included in "any-word".
>
>
'Ang on. If zo-word is ZO+any-word, and and any-word can be a zo-word,
then doesn't the grammar imply that "zo zo broda" is *one* word, a
doubly-quoted "broda"? That seems to be the precedence we get from these.


~mark


posts: 1912


> Robin Lee Powell scripsit:
>
> > {broda ba'e si broda} does very much the wrong thing under the
> > Exceptionless Rules, for example; the result is {broda *si* broda},
> > so si stays in the speech stream.
>
> Hmm. I read the rules as generating {broda} (the first broda, that is),

(The first broda is erased and the second broda remains.)

> as while ba'e tags the following word as emphatic, it does not
> prevent it from having whatever effect it has.

Exactly.

> Which means you can't erase a BAhE with "si". Which sucks.

You can erase with "any-word-but-BAhE SI", for example
"ba'e da si" will erase any traces of {ba'e}.

I think that little inconvenience is much preferrable to
the introduction of exceptions.

> BTW, it should say somewhere that after erasure words have done their
> work, any remaining erasure words make the utterance ungrammatical.

Or, they could be allowed at the beginning of text. Those would
be exceptional cases, but the beginning of text is already full
of exceptions anyway.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> >zo-word = ZO any-word
> >bu-word = any-word BU
> >
> >Where:
> >
> >any-word = zo-word | zei-word | bu-word | lo'u-word
> > | zoi-word | A | ... | ZOhU
> >
> 'Ang on. If zo-word is ZO+any-word, and and any-word can be a zo-word,
> then doesn't the grammar imply that "zo zo broda" is *one* word, a
> doubly-quoted "broda"? That seems to be the precedence we get from these.

If I understand correctly how this works, in {zo zo broda} first {zo zo}
will be processed as a "zo-word", and then we have "zo-word broda". The
second zo never gets a chance to form a zo-word.

OTOH, in {zo broda bu}, {zo broda} first becomes a "zo-word", and
then {zo-word bu} becomes a "bu-word".

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




Jorge Llamb�as scripsit:

> > Which means you can't erase a BAhE with "si". Which sucks.
>
> You can erase with "any-word-but-BAhE SI", for example
> "ba'e da si" will erase any traces of {ba'e}.
>
> I think that little inconvenience is much preferrable to
> the introduction of exceptions.

Exceptions are all in how you look at it. The naive explanation of "si"
is "erases the word in front of it". Under Jorge's Exceptionless Rules
(as distinct from some other set of exceptionless rules, and it seems
strange to call rules "exceptionless" when each begins with "Unless",
but hey, whatever) it's more like "erases the word in front of it,
except when ... and when ... and when ..."

Simplicity, too, is all in how you look at it.

--
Unless it was by accident that I had John Cowan
offended someone, I never apologized. jcowan@reutershealth.com
--Quentin Crisp http://www.ccil.org/~cowan


posts: 1912


>
> Exceptions are all in how you look at it.

Agreed.

> The naive explanation of "si"
> is "erases the word in front of it".

Right. But in Lojban, normally, everything is processed
on a first come first served basis. That's the meta-rule
I'm starting with. Then in {ba'e si}, ba'e is served first,
and then this emphatic si, considered as a single thing,
does erase the word in front of it.

> Under Jorge's Exceptionless Rules
> (as distinct from some other set of exceptionless rules, and it seems
> strange to call rules "exceptionless" when each begins with "Unless",
> but hey, whatever)

That's probably not the ideal wording, agreed.

> it's more like "erases the word in front of it,
> except when ... and when ... and when ..."

When we are describing the word in isolation, that's true.
When we describe how the whole thing works together, they
are not exceptions.

> Simplicity, too, is all in how you look at it.

Yes, I agree. I find the "first come first served" meta-rule
the simplest among the alternatives considered so far.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




Robin Lee Powell scripsit:

> You mean happy about the lerfu formed by A+BU, but not the BU,
> correct?

Well, most of us aren't happy about *letters*. So in Mark~'s
example, it's Mr. A that he is happy about, not the letter "a",
still less the Lojban word "bu".

--
In politics, obedience and support John Cowan <jcowan@reutershealth.com>
are the same thing. --Hannah Arendt http://www.ccil.org/~cowan


posts: 1912


> Robin Lee Powell scripsit:
>
> > You mean happy about the lerfu formed by A+BU, but not the BU,
> > correct?
>
> Well, most of us aren't happy about *letters*. So in Mark~'s
> example, it's Mr. A that he is happy about, not the letter "a",
> still less the Lojban word "bu".

The same thing happens with {le broda ku ui}. That doesn't
express happiness about the word {ku} or the sumti {le broda ku},
but about the referent of the sumti.

I think that it doesn't matter much if the grammar attaches
the UI to {bu} or to {abu}. It makes no difference, just as
it would make no difference attaching it to {ku} or to
{le broda ku} as a whole.

In the case of {zo a ui} however, I think it makes more sense
to attach {ui} to the zo-word as a whole, i.e., ui is not quoted
here.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 2388



>it seems
> strange to call rules "exceptionless" when each
> begins with "Unless",
>

There is a principle (useful in ethics, for
example) to distinguish between rules with
exxceptions and conditional rules. A conditional
rules says "If ... then ---," a rule with
exceptions say "---" but then doesn't work right
in some cases, which may or may not be noted.
Since it is the note that is important here, what
is wanted is conditional rules, not rules with
exceptions. (Kant demanded exceptionless rules
but forgot — usually, anyhow — to require
categorical ones, i.e., non-conditional).


On Sun, 7 Nov 2004, Robin Lee Powell wrote:

>> I very much like option 3. I'm curious about one point you
>> mentioned earlier, and whether it's still a problem, though.
>> After erasing the final delimiter, are we in a place to add more
>> quoted text?
>
> Yes, absolutely.

Hmmm. When I wrote that yesterday, I thought that was a good thing. Now I'm
confused with how that interacts with the "four {si}s remove a {zoi}" thing.
That is, consider

zoi .kor. text .kor. si si si si

Wouldn't the first si remove the last .kor., and then the rest all fall into
the quotes, the same as if you'd said

zoi .kor. text si si si

and fail, because there's no closing delimiter? That is, how do we know that
the next si is meant to remove the zoiWords, and not be added to them?

>
>> ISTR you saying that
>>
>> zoi zoi I'm talking zoi si about quotes zoi
>>
>> was invalid, or at least different from
>>
>> zoi zoi I'm talking about quotes zoi
>>
>> because although the si ate the terminating zoi, the first zoi had
>> already turned it into something else and stopped working.
>
> That was when I was using a pre-processor. In the default mode, my
> current parser outputs the following for the first one:
>
> text
> zoiClause
> |- ZOI: zoi
> |- anyWordZoi: zoi
> |- zoiPrintableWord: I'm
> |- zoiWords
> |- zoiPrintableWord: talking
> |- zoiWords
> |- zoiPrintableWord: about
> |- zoiWords
> |- zoiPrintableWord: quotes
> |- anyWordZoi: zoi
>
>> Thanks for trying to comprehend all these interactions. I'm just
>> barely able to keep my head from exploding on many of them.
>
> Thanks for the encouragement. :-)
>
> xorxes is working hard on this too. Thanks, xorxes.

..i ki'esai .xorxes.
--
Adam Lopresto
http://cec.wustl.edu/~adam/

Aleph-null bottles of beer on the wall,
Aleph-null bottles of beer,
You take one down, and pass it around,
Aleph-null bottles of beer on the wall.


posts: 14214

On Mon, Nov 08, 2004 at 07:00:24AM -0800, Jorge Llamb?as wrote:
> > it's more like "erases the word in front of it, except when ...
> > and when ... and when ..."
>
> When we are describing the word in isolation, that's true. When we
> describe how the whole thing works together, they are not
> exceptions.
>
> > Simplicity, too, is all in how you look at it.
>
> Yes, I agree. I find the "first come first served" meta-rule the
> simplest among the alternatives considered so far.

OK, we have a conflation problem.

xorxes' proposal introduces *two* rules: left-to-right ordering,

  • and* consumption of clauses rather than words.


John was talking about the latter, but xorxes responded as though he
was talking about the former.

One way to look at xorxes' SI is "SI consumes words, or clauses
formed by other magic words".

The other way (and, in fact, the much more likely way for someone
who hasn't internalized formalized grammars to see it) is "SI
consumes words, or zo+word, or zoi+delim+stuff+delim, or word+bu,
or, or, or, or...".

This was John's point; left-to-right ordering has nothing to do with
it.

-Robin


posts: 14214

On Mon, Nov 08, 2004 at 01:00:05PM -0600, Adam D. Lopresto wrote:
> On Sun, 7 Nov 2004, Robin Lee Powell wrote:
>
> >>I very much like option 3. I'm curious about one point you
> >>mentioned earlier, and whether it's still a problem, though.
> >>After erasing the final delimiter, are we in a place to add more
> >>quoted text?
> >
> >Yes, absolutely.
>
> Hmmm. When I wrote that yesterday, I thought that was a good
> thing. Now I'm confused with how that interacts with the "four
> {si}s remove a {zoi}" thing.

Strings of SI are excepted. The interactions at that point become
hairy if you don't also disallow SI as a delimiter.

> That is, how do we know that the next si is meant to remove the
> zoiWords, and not be added to them?

We define that you can't do the latter. You can, however, say:

zoi zoi foo zoi si <anything else here> si si si

and put si in that way.

-Robin


posts: 14214

On Mon, Nov 08, 2004 at 08:32:12AM -0800, Jorge Llamb?as wrote:
> In the case of {zo a ui} however, I think it makes more sense to
> attach {ui} to the zo-word as a whole, i.e., ui is not quoted
> here.

I'm confused. How could .ui possibly be quoted there?

-Robin


posts: 14214

On Mon, Nov 08, 2004 at 06:20:15AM -0800, Jorge Llamb?as wrote:
> I think that little inconvenience is much preferrable to the
> introduction of exceptions.

See, that right there may be a sticking point between us.

-Robin


posts: 14214

On Mon, Nov 08, 2004 at 06:20:15AM -0800, Jorge Llamb?as wrote:
> > BTW, it should say somewhere that after erasure words have done
> > their work, any remaining erasure words make the utterance
> > ungrammatical.
>
> Or, they could be allowed at the beginning of text. Those would be
> exceptional cases, but the beginning of text is already full of
> exceptions anyway.

Believe me, if you don't allow erasures at the beginning of text,
parsing IRC is gonna *SUCK*.

-Robin


posts: 14214

On Mon, Nov 08, 2004 at 05:56:17AM -0800, Jorge Llamb?as wrote:
snip
> SA has to be handled selmaho by selmaho.

Oh god, does it ever.

The PEG grammar is ~1700 lines, including comments and blank lines.

SA handling is, approximately, 600 of those lines...

-Robin



posts: 14214

On Mon, Nov 08, 2004 at 05:56:17AM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > In fact, ba'e will cause many problems, even with left-to-right
> > ordering, which is why I made a seperate section ("Marking
> > Words") for it and UI.
>
> That's an important point.

I thought so.

> If we define "magic word" as any word that deals with the
> construct "any-word", then UI and BAhE are not magic words. They
> do deal with every word in the same way, but they don't deal with
> the construct "any-word" or anything like it, which disables the
> ordinary function of any word.

<nod>

> zoi-word =zoi any-word anything that-word
> (I'm not sure how this one is formalized)

Poorly; truly formalizing it requires a full-on Context Sensitive
Grammar. Thanks, but no thanks. My parser uses code attached to
the production to do it, and I assume the other two are the same.

> Where:
>
> any-word = zo-word | zei-word | bu-word | lo'u-word
> | zoi-word | A | ... | ZOhU

Well, in your universe. Don't forget BRIVLA and CMENE.

> > {broda ba'e si broda} does very much the wrong thing under the
> > Exceptionless Rules, for example; the result is {broda *si*
> > broda}, so si stays in the speech stream.
>
> {broda ba'e si brode} would be an emphatically deleted {broda},
> replaced by {brode}: i.e. {ba'e} is always (no exceptions)
> absorbed by the following word. The construct {ba'e si} would be
> an emphasized word eraser.

I'm sorry, but no matter how many times I read that my brain still
treats it as an exception. SI is just *more* *important* then BAhE
to me, and it seems bizarre that SI wouldn't do its proper job
(erasing BAhE) in this case.

However, someone who "grew up" with your system likely wouldn't feel
that way. I just wanted to get it out.

-Robin


posts: 1912


> xorxes' proposal introduces *two* rules: left-to-right ordering,
> *and* consumption of clauses rather than words.

They are independent. You could do without the second one, all
you would have is more strings resulting in error. In other
words, the second rule is there only to handle what would
otherwise be error cases (failure to find a suitable word).

The only cmavo affected by this are {bu}, {zei} and {si}.

> One way to look at xorxes' SI is "SI consumes words, or clauses
> formed by other magic words".

Right.

> The other way (and, in fact, the much more likely way for someone
> who hasn't internalized formalized grammars to see it) is "SI
> consumes words, or zo+word, or zoi+delim+stuff+delim, or word+bu,
> or, or, or, or...".
>
> This was John's point; left-to-right ordering has nothing to do with
> it.

It has something to do with it. There would be no problem with
allowing SI to eat just the last word of the magic word clause,
but then, by left-to-right ordering the magic word is left
wanting a word, so a second SI will be gobbled by the magic
word. At this point is where you have to start to give
right-to-left precedence if you don't want this to happen.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> On Mon, Nov 08, 2004 at 08:32:12AM -0800, Jorge Llamb?as wrote:
> > In the case of {zo a ui} however, I think it makes more sense to
> > attach {ui} to the zo-word as a whole, i.e., ui is not quoted
> > here.
>
> I'm confused. How could .ui possibly be quoted there?

I mean that {zo a ui} is equivalent to {lu a li'u ui},
not to {lu a ui li'u}.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> > any-word = zo-word | zei-word | bu-word | lo'u-word
> > | zoi-word | A | ... | ZOhU
>
> Well, in your universe. Don't forget BRIVLA and CMENE.

They are included in the ... :-)

> > {broda ba'e si brode} would be an emphatically deleted {broda},
> > replaced by {brode}: i.e. {ba'e} is always (no exceptions)
> > absorbed by the following word. The construct {ba'e si} would be
> > an emphasized word eraser.
>
> I'm sorry, but no matter how many times I read that my brain still
> treats it as an exception. SI is just *more* *important* then BAhE
> to me, and it seems bizarre that SI wouldn't do its proper job
> (erasing BAhE) in this case.

If you say that SI has priority over everything, then it is an exception.
If you say that left-to-right processing has priority, then it is not
an exception.

But you don't consider SI more important than everything: ZO is
more important than SI, at least more so than the first SI.
But then three SI's together can defeat a ZO.

Having to remember the order of precedence of all the magic words,
plus all the exceptions and counter-exceptions to that order of
precedence, is too much trouble for what it buys us, in my opinion.

> However, someone who "grew up" with your system likely wouldn't feel
> that way. I just wanted to get it out.

I don't know why you feel that SI is more important than BAhE
but less important than ZO, other than because you have become used
to it.

I have not been able to learn the official precedence rules
of magic words in about 10 years of doing Lojban. It is not even
clear if there is an official set of consistent and complete
rules. I just hope we don't end up with another unlearnable set.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> Oh, another thing: in left-to-right processing, BU *always* loses.
> Not sure this is a bad thing, but wanted to point it out.

"Always" here means {zo bu}, {zoi bu ... bu}, {da zei bu},
{lo'u bu le'u}, {da ba'e bu}, {fa'o bu}.

In how many of those does {bu} win with your rules? In fact,
does it win in any one at all?

But there is always a way to get a lerfu from those words:
{zo zo bu}, {zo zoi bu}, {zo zei bu}, {zo lo'u bu},
{zo le'u bu}, {zo ba'e bu}, {zo fa'o bu} all give a lerfu
from the quoted word.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


I inadvertently left FAhO out when doing the formal rules
for the magic words in "exceptionless rules" mode.
Would the following do what I want for FAhO?

text = (rest of the grammar) faho-stop

faho-stop = FAhO / any-word faho-stop

What I would want is for FAhO to absorb any extra word
as needed to leave a grammatical text behind. Is that
how FAhO is supposed to work?

mu'o mi'e xorxes

posts: 1912


Some more comments.

> ZO quotes the following word, no matter what it is, except Y.
> Words quoted with ZO lose their grammatical functions.

This doesn't mention {zo zo da} and {zoi zo da ........ zo de}.

If we can't assume left-to-right precedence, and {zo}
wins over everything else, then in those cases {da} and {de}
would be quoted by {zo}. Also the difference between
{zo da si de} and {zo da si si} should be mentioned here.

> FAhO terminates the word or text stream unequivocably,
> unless quoted with ZO.

Even within {lo'u ... le'u}?

> LOhU quotes all following Lojban words up to a LEhU (but
> not a ZO+LEhU; this is to allow nested LOhU...LEhU quotes
> inside a LOhU...LEhU, so you can talk about mistakes you
> made that include a previous error quote). Except for the
> ZO+LEhU case (which should be read as simply LEhU inside
> a LOhU...LEhU quote) all Lojban words within a LOhU...LEhU
> quote are read without any grammar and have no
> grammatical effects.

Even FAhO? If FAhO is active within LOhU-LEhU, what about
ZO+FAhO?

> LEhU is ungrammatical except at the end of a LOhU quotation
> and after ZO.

Is {le'u si} ungrammatical?

> ZOI cmavo use the following word as a delimiting word, no
> matter what it is, execept Y (which is ignored); and SI, SA
> and SU (which erase it).

Can it use ZO, FAhO, LEhU?

> BU makes the preceding word into a lerfu word, except for ZO,
> ZEI and LOhU (which quote it), ZOI (which uses it as a delimiter);
> LEhU (which would result in bizarrely re-opening the LOhU...LEhU,
> and hence is an error); SI, SA and SU (it affects whatever is to the
> left after the erasing is done; if nothing is left an error results);
> BAhE (which it skips); BU (which would lead to grouping issues,
> and hence is an error to attempt); and FAhO (after which anything,
> including BU, is ignored).

What about {zo da bu} and {da zei de bu}?

mu'o mi'e xorxes

posts: 14214

A simple question:

How many of you think that the rules as I currently have them written would be very hard for you to learn?

For those of you who have also read xorxes' stuff, do you find it easier?

-Robin

posts: 14214

On Tue, Nov 09, 2004 at 08:14:00AM -0800, wikidiscuss@lojban.org
wrote:
> Re: Magic Words
>
> I inadvertently left FAhO out when doing the formal rules for the
> magic words in "exceptionless rules" mode. Would the following do
> what I want for FAhO?
>
> text = (rest of the grammar) faho-stop
>
> faho-stop = FAhO / any-word faho-stop
>
> What I would want is for FAhO to absorb any extra word as needed
> to leave a grammatical text behind. Is that how FAhO is supposed
> to work?

No. fa'o is a complete cessation of input; if what leads up to the
fa'o isn't grammatical, then an error results.

-Robin


posts: 14214

On Tue, Nov 09, 2004 at 10:55:37AM -0800, wikidiscuss@lojban.org
wrote:
> Re: Magic Words
>
> Some more comments.
>
> > ZO quotes the following word, no matter what it is, except Y.
> > Words quoted with ZO lose their grammatical functions.
>
> This doesn't mention {zo zo da} and {zoi zo da ........ zo de}.
>
> If we can't assume left-to-right precedence,

Do you mean top-to-bottom?

> and {zo} wins over everything else, then in those cases {da} and
> {de} would be quoted by {zo}.

That's true in the second case, but not the first.

> > FAhO terminates the word or text stream unequivocably, unless
> > quoted with ZO.
>
> Even within {lo'u ... le'u}?

Err, no. Sorry. Nor zoi, for that matter.

> > LEhU is ungrammatical except at the end of a LOhU quotation and
> > after ZO.
>
> Is {le'u si} ungrammatical?

Obviously not, I should think.

> > ZOI cmavo use the following word as a delimiting word, no matter
> > what it is, execept Y (which is ignored); and SI, SA and SU
> > (which erase it).
>
> Can it use ZO, FAhO, LEhU?

I don't see why not. As mentioned in another mail, zoi needs to be
moved higher up.

> > BU makes the preceding word into a lerfu word, except for ZO,
> > ZEI and LOhU (which quote it), ZOI (which uses it as a
> > delimiter); LEhU (which would result in bizarrely re-opening the
> > LOhU...LEhU, and hence is an error); SI, SA and SU (it affects
> > whatever is to the left after the erasing is done; if nothing is
> > left an error results); BAhE (which it skips); BU (which would
> > lead to grouping issues, and hence is an error to attempt); and
> > FAhO (after which anything, including BU, is ignored).
>
> What about {zo da bu} and {da zei de bu}?

Handled by the meta-rules; both are errors.

-Robin


posts: 14214

On Mon, Nov 08, 2004 at 03:20:20PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > Oh, another thing: in left-to-right processing, BU *always*
> > loses. Not sure this is a bad thing, but wanted to point it out.
>
> "Always" here means {zo bu}, {zoi bu ... bu}, {da zei bu}, {lo'u
> bu le'u}, {da ba'e bu}, {fa'o bu}.

Also "broda si bu", actually.

> In how many of those does {bu} win with your rules? In fact, does
> it win in any one at all?

Nope, don't think so. Just wanted to point it out, is all.

> But there is always a way to get a lerfu from those words: {zo zo
> bu}, {zo zoi bu}, {zo zei bu}, {zo lo'u bu}, {zo le'u bu}, {zo
> ba'e bu}, {zo fa'o bu} all give a lerfu from the quoted word.

<nod>

Would you mind collecting your version, clearly labelling it as
such, and putting it at the bottom of the Magic Words page?

It's certainly worthy enough of consideration that people should
have the two versions side-by-side to compare. I'll mark it up with
example comparisons and such.

-Robin


posts: 1912


> Would you mind collecting your version, clearly labelling it as
> such, and putting it at the bottom of the Magic Words page?
>
> It's certainly worthy enough of consideration that people should
> have the two versions side-by-side to compare. I'll mark it up with
> example comparisons and such.

vi'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> > text = (rest of the grammar) faho-stop
> >
> > faho-stop = FAhO / any-word faho-stop
> >
> > What I would want is for FAhO to absorb any extra word as needed
> > to leave a grammatical text behind. Is that how FAhO is supposed
> > to work?
>
> No. fa'o is a complete cessation of input; if what leads up to the
> fa'o isn't grammatical, then an error results.

Wouldn't it be useful to be able to bail out of a text in
midsentence without losing any grammatical prior parts?

If that's not the case, then FAhO is not a magic word, since
it doesn't deal with any-word in any way.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Tue, Nov 09, 2004 at 01:15:32PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > > text = (rest of the grammar) faho-stop
> > >
> > > faho-stop = FAhO / any-word faho-stop
> > >
> > > What I would want is for FAhO to absorb any extra word as
> > > needed to leave a grammatical text behind. Is that how FAhO is
> > > supposed to work?
> >
> > No. fa'o is a complete cessation of input; if what leads up to
> > the fa'o isn't grammatical, then an error results.
>
> Wouldn't it be useful to be able to bail out of a text in
> midsentence without losing any grammatical prior parts?

Heck no, because then you can't say anything else!

> If that's not the case, then FAhO is not a magic word, since it
> doesn't deal with any-word in any way.

Sure it does:

faho-stop = FAhO any-word*

I actually use .* instead in my parser, though.

-Robin


On Tue, 9 Nov 2004 wikidiscuss@lojban.org wrote:

> Re: Magic Words
> A simple question:
>
> How many of you think that the rules as I currently have them written would
> be very hard for you to learn?

For the most part, I think they're not too bad, but there are a few parts that
I know I'd have trouble with. I think the "zo le'u inside lo'u" stuff in
particular comes across as something of a wart (the number of times it gets
mentioned, for instance, as exceptions to other rules, makes it annoying).

Also, I might mention here an issue I've been wanting to post elsewhere, you've
got

BU makes the preceding word into a lerfu word, except for ZO, ... BU (which
would lead to grouping issues, and hence is an error to attempt); ... Multiple
BU may be used in succession, in which case a new letteral is formed for each
additional BU (i.e. "broda bu" is a different letteral from "broda bu bu").
However, "bu bu" by itself is illegal.

That seems to contradict itself about bu bu. Should the "BU (which would lead
to grouping issues, and hence is an error to attempt);" part be stricken?

Also, I was led to wonder whether "fa'o si" is legal. I'm sure in normal
conversation only the biggest pedant would stick his fingers in his ears and
ignore "fa'o si fo'a", but is it actually valid?

Also, you have "SU+SI == nothing". Does that mean nothing in the sense of
"text before the SU is undisturbed" or nothing as in "no text is left
whatsoever". What does "SA+SU" or "SA+SA" mean? Is SA an error if there are
no previous words of the desired selma'o?

> For those of you who have also read xorxes' stuff, do you find it easier?

Easier to learn, absolutely, no question about it. Easier to use, not really,
since yours make it easier to, for instance, make corrections piecewise in zoi
delimiters. I'd really like to see actual texts where there are differences in
interpretation between the two sets of rules, to get an idea of how often it
matters. I'm tempted to say if the interactions are as rare as I suspect, we
might be better with the simpler rules, since not much will be lost. But I'm
still conflicted.
--
Adam Lopresto
http://cec.wustl.edu/~adam/

MacGyver is the Martha Stewart of action.
--Patrick J. Mooney


posts: 14214

On Tue, Nov 09, 2004 at 04:32:00PM -0600, Adam D. Lopresto wrote:
> On Tue, 9 Nov 2004 wikidiscuss@lojban.org wrote:
>
> >Re: Magic Words A simple question:
> >
> >How many of you think that the rules as I currently have them
> >written would be very hard for you to learn?
>
> For the most part, I think they're not too bad, but there are a
> few parts that I know I'd have trouble with. I think the "zo le'u
> inside lo'u" stuff in particular comes across as something of a
> wart (the number of times it gets mentioned, for instance, as
> exceptions to other rules, makes it annoying).

Heh. It *is* part of the current language, though, and removing it
constitutes a change to the language as it stands.

> Also, I might mention here an issue I've been wanting to post
> elsewhere, you've got
>
> BU makes the preceding word into a lerfu word, except for ZO, ...
> BU (which would lead to grouping issues, and hence is an error to
> attempt); ... Multiple BU may be used in succession, in which case
> a new letteral is formed for each additional BU (i.e. "broda bu"
> is a different letteral from "broda bu bu"). However, "bu bu" by
> itself is illegal.
>
> That seems to contradict itself about bu bu. Should the "BU
> (which would lead to grouping issues, and hence is an error to
> attempt);" part be stricken?

"bu bu" by itself is an error. "da bu bu" is not.

> Also, I was led to wonder whether "fa'o si" is legal.

  • Anything* after fa'o is legal. In both versions, though, the text

stream ends at the fa'o, and si can't fix it.

> Also, you have "SU+SI == nothing". Does that mean nothing in the
> sense of "text before the SU is undisturbed" or nothing as in "no
> text is left whatsoever".

The former.

> What does "SA+SU" or "SA+SA" mean?

Both are errors.

> Is SA an error if there are no previous words of the desired
> selma'o?

Currently, yes, although in my parser I actually allow SA at the
beginning of text, and I'd want that in whatever we settle on.

> >For those of you who have also read xorxes' stuff, do you find it
> >easier?
>
> Easier to learn, absolutely, no question about it.

OK.

> Easier to use, not really, since yours make it easier to, for
> instance, make corrections piecewise in zoi delimiters.

<nod>

> I'd really like to see actual texts where there are differences in
> interpretation between the two sets of rules, to get an idea of
> how often it matters.

Ask, and ye shall receive.

http://www.lojban.org/tiki//tiki-index.php?page=Magic+Words%3A+left-to-right

Made a big comparison table at the bottom.

-Robin


On Tuesday 09 November 2004 11:14, wikidiscuss@lojban.org wrote:
> Re: Magic Words
>
> I inadvertently left FAhO out when doing the formal rules
> for the magic words in "exceptionless rules" mode.
> Would the following do what I want for FAhO?
>
> text = (rest of the grammar) faho-stop
>
> faho-stop = FAhO / any-word faho-stop
>
> What I would want is for FAhO to absorb any extra word
> as needed to leave a grammatical text behind. Is that
> how FAhO is supposed to work?

{fa'o} can be followed by anything, whether it's valid Lojban words or not,
and still be grammatical.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 84

wikidiscuss@lojban.org (Robin) wrote:

>Re: Magic Words
>A simple question:
>
>How many of you think that the rules as I currently have them written would be very hard for you to learn?
>
>For those of you who have also read xorxes' stuff, do you find it easier?
>
>
xorxes is frequently over my head, but I have to admit that I can see
his way pretty clearly. The left-to-right principle has only the
advantage of simplicity over precedence rules, and precedence rules
aren't that tough, so I can see either way; the concept of "magic-word
output", though, really does make sense to me and makes things a whole
lot simpler in analyzing the interactions involved.

~mark



posts: 84

Jorge "Llamb����������������������������������" wrote:

(who?)

>--- Robin Lee Powell wrote:
>
>
>>>text = (rest of the grammar) faho-stop
>>>
>>>faho-stop = FAhO / any-word faho-stop
>>>
>>>What I would want is for FAhO to absorb any extra word as needed
>>>to leave a grammatical text behind. Is that how FAhO is supposed
>>>to work?
>>>
>>>
>>No. fa'o is a complete cessation of input; if what leads up to the
>>fa'o isn't grammatical, then an error results.
>>
>>
>
>Wouldn't it be useful to be able to bail out of a text in
>midsentence without losing any grammatical prior parts?
>
>If that's not the case, then FAhO is not a magic word, since
>it doesn't deal with any-word in any way.
>
>
It's still magic, but in a different way. FAhO is equivalent to
"silence forever on this input stream". What's important is less its
effect on following words (there are no following words; they don't
count or are a different stream), but under what circumstances is its
own functioning cancelled. Since it does affect the whole input and
parsing routine, this is more "magical" than working out when the
function of LE or BAI is cancelled.

~mark



posts: 84

Adam D. Lopresto wrote:

>> For those of you who have also read xorxes' stuff, do you find it
>> easier?
>
>
> Easier to learn, absolutely, no question about it. Easier to use, not
> really,
> since yours make it easier to, for instance, make corrections
> piecewise in zoi
> delimiters.

I'm thinking that making corrections piecewise in zoi delimiters is not
something we want to enable. Correcting piecewise inside LOhU/LEhU
maybe, but zoi is powerful juju.

~mark



posts: 1912


> FAhO is equivalent to
> "silence forever on this input stream". What's important is less its
> effect on following words (there are no following words; they don't
> count or are a different stream), but under what circumstances is its
> own functioning cancelled.

In left-to-right, it will be cancelled whenever there is a preceding
word that grabs it as "any-word". Those cases are: {zo fa'o},
{lo'u ... fa'o ...le'u}, {zoi fa'o .... fa'o}, and {da zei fa'o}.

Making {fa'o} not count as "any-word" in any one of those
would constitute an exception to left-to-right processing.

If {fa'o} does not allow reading any following words, then
{fa'o bu}, {fa'o zei da}, {fa'o si}, {fa'o sa ...}, {fa'o su}
are all impossible: {fa'o} is active here and there cannot be
any following words to parse.

{ba'e fa'o} is not problematic: an emphatic end-of-text.
{sa fa'o} is not problematic either: it will delete everything
because it won't find any other word from selmaho FAhO, and
then {fa'o} itself will end text. It has the same effect as
{su fa'o}.

> Since it does affect the whole input and
> parsing routine, this is more "magical" than working out when the
> function of LE or BAI is cancelled.

Yes, it's much more powerful once it can fulfill its function, but
that doesn't seem like a reason to introduce exceptions to allow it
to fulfill its function in places where no other word is allowed to.

There are other issues I have with FAhO, though.

1) If we really need such a metagrammatical word for
end-of-parsable-input, how come a corresponding word for
beginning-of-parsable-input is not needed?

2) At least in Robin's parser implementation, FAhO has a true
magic word grammatical function besides its end-of-parsable-input
metagrammatical function. If I understand correctly, in Robin's
parser FAhO swallows also any preceding word that makes the
text ungrammatical. So for example: {mi klama le zarci le fa'o}
will be parsed as {mi klama le zarci /ku/ /vau/ fa'o}. So FAhO
is also in the SI, SA, SU group in that it erases zero or more
preceding words.

3) I think it would be better if FAhO was just an "end of
grammatical text" (elidable) terminator, i.e.:
text = (rest of grammar) /FAhO/
This way it would not be a magic word, it could be erased
with SI/SA/SU, and used as "any-word" by BU and ZEI.
So you could easily fix a {fa'o si fo'a} mistake.

Recognizing end-of-parsable-input is a metagrammatical
function, akin to recognizing beginning-of-parsable-input.
Doing it with a word that can also be used within the
grammar (as in {zo fa'o}) is weird.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




On Wed, 10 Nov 2004, Jorge Llambas wrote:

> There are other issues I have with FAhO, though.
>
> 1) If we really need such a metagrammatical word for
> end-of-parsable-input, how come a corresponding word for
> beginning-of-parsable-input is not needed?

I think {su} serves that purpose. Anything at all is allowed before it, and
ignored as far as the parser is concerned. Or is that not the case?

> 2) At least in Robin's parser implementation, FAhO has a true
> magic word grammatical function besides its end-of-parsable-input
> metagrammatical function. If I understand correctly, in Robin's
> parser FAhO swallows also any preceding word that makes the
> text ungrammatical. So for example: {mi klama le zarci le fa'o}
> will be parsed as {mi klama le zarci /ku/ /vau/ fa'o}. So FAhO
> is also in the SI, SA, SU group in that it erases zero or more
> preceding words.
>
> 3) I think it would be better if FAhO was just an "end of
> grammatical text" (elidable) terminator, i.e.:
> text = (rest of grammar) /FAhO/
> This way it would not be a magic word, it could be erased
> with SI/SA/SU, and used as "any-word" by BU and ZEI.
> So you could easily fix a {fa'o si fo'a} mistake.
>
> Recognizing end-of-parsable-input is a metagrammatical
> function, akin to recognizing beginning-of-parsable-input.
> Doing it with a word that can also be used within the
> grammar (as in {zo fa'o}) is weird.
>
> mu'o mi'e xorxes
>
>
>
>
> __
> Do you Yahoo!?
> Check out the new Yahoo! Front Page.
> www.yahoo.com
>
>
>
>

--
Adam Lopresto
http://cec.wustl.edu/~adam/

"Yet creeds mean very little," Coth answered the dark god, still speaking
almost gently. "The optimist proclaims that we live in the best of all
possible worlds; and the pessimist fears this is true."

--James Cabell, "The Silver Stallion."

Jorge Llambías wrote:
> In left-to-right, it will be cancelled whenever there is a preceding
> word that grabs it as "any-word". Those cases are: {zo fa'o},
> {lo'u ... fa'o ...le'u}, {zoi fa'o .... fa'o}, and {da zei fa'o}.

What about {zoi zoi ... fa'o ... zoi}?

mu'o mi'e .filip.


posts: 1912



> Jorge Llambías wrote:
> > In left-to-right, it will be cancelled whenever there is a preceding
> > word that grabs it as "any-word". Those cases are: {zo fa'o},
> > {lo'u ... fa'o ...le'u}, {zoi fa'o .... fa'o}, and {da zei fa'o}.
>
> What about {zoi zoi ... fa'o ... zoi}?

Anything (except ".zoi." in this case) can go there, it's not even
identified as a word. It is just a part of the foreign text. It is
not grabbed as "any-word" but as a part of "anything".

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 1912


> On Wed, 10 Nov 2004, Jorge Llambías wrote:
> > 1) If we really need such a metagrammatical word for
> > end-of-parsable-input, how come a corresponding word for
> > beginning-of-parsable-input is not needed?
>
> I think {su} serves that purpose. Anything at all is allowed before it, and
> ignored as far as the parser is concerned. Or is that not the case?

Sort of, but I think it will only delete Lojban words, at least if
its grammar really is:

text = su-null (rest of grammar)
su-null = SU / any-word su-null

But I don't really know if that is its grammar, it is not explicited
anywhere.

In any case, the way I understand SU, the parser will process
everything and only when it gets to a valid SU erase everything.
A true "beginning-of-parsable-input" could not be quoted for
example ({zo su} would be impossible), in the same way that
{fa'o bu} is impossible, because no processing would occur outside
of those markers.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Wed, Nov 10, 2004 at 08:27:10AM -0500, Mark E. Shoulson wrote:
> wikidiscuss@lojban.org (Robin) wrote:
>
> >Re: Magic Words A simple question:
> >
> >How many of you think that the rules as I currently have them
> >written would be very hard for you to learn?
> >
> >For those of you who have also read xorxes' stuff, do you find it
> >easier?
>
> xorxes is frequently over my head, but I have to admit that I can
> see his way pretty clearly. The left-to-right principle has only
> the advantage of simplicity over precedence rules, and precedence
> rules aren't that tough,

Heh. They will be when Raphael is done with them...

-Robin


posts: 14214

On Wed, Nov 10, 2004 at 08:43:03AM -0800, Jorge Llamb?as wrote:
> --- "Mark E. Shoulson" wrote:
> > Since it does affect the whole input and parsing routine, this
> > is more "magical" than working out when the function of LE or
> > BAI is cancelled.
>
> Yes, it's much more powerful once it can fulfill its function, but
> that doesn't seem like a reason to introduce exceptions to allow
> it to fulfill its function in places where no other word is
> allowed to.
>
> There are other issues I have with FAhO, though.
>
> 1) If we really need such a metagrammatical word for
> end-of-parsable-input, how come a corresponding word for
> beginning-of-parsable-input is not needed?

Because we couldn't process it: the preceeding text, not being valid
Lojban (by definiton) can't be parsed, and probably can't even be
broken up syntactically. Furthermore, whatever word we chose, it
could appear in the preceeding (non-Lojban) text in error.

> 2) At least in Robin's parser implementation, FAhO has a true
> magic word grammatical function besides its end-of-parsable-input
> metagrammatical function. If I understand correctly, in Robin's
> parser FAhO swallows also any preceding word that makes the
> text ungrammatical.

  • NO*. This is *not* correct.


PEG parsers, by default, output only those things they processed
successfully. A grammatical error before fa'o causes that error and
all that follows, *including* fa'o, to not be processed. Sort of.
It's really more complicated than that.

So the effect you're assigning to fa'o actually happens no matter
what, fa'o or not.

> 3) I think it would be better if FAhO was just an "end of
> grammatical text" (elidable) terminator, i.e.: text = (rest of
> grammar) /FAhO/ This way it would not be a magic word, it could be
> erased with SI/SA/SU, and used as "any-word" by BU and ZEI. So you
> could easily fix a {fa'o si fo'a} mistake.

This removes the primary function of fa'o, which is to allow
arbitrary non-Lojban text afterwards.

Think of it as an uncloseable zoi.

Whether that function is important or not is a seperate discussion.

> Recognizing end-of-parsable-input is a metagrammatical function,
> akin to recognizing beginning-of-parsable-input. Doing it with a
> word that can also be used within the grammar (as in {zo fa'o}) is
> weird.

Now *that* I wholeheartedly agree with. fa'o is wierd. Especially
since we have mu'o.

-Robin


posts: 14214

On Wed, Nov 10, 2004 at 10:58:07AM -0600, Adam D. Lopresto wrote:
> On Wed, 10 Nov 2004, Jorge Llamb?as wrote:
>
> >There are other issues I have with FAhO, though.
> >
> >1) If we really need such a metagrammatical word for
> >end-of-parsable-input, how come a corresponding word for
> >beginning-of-parsable-input is not needed?
>
> I think {su} serves that purpose. Anything at all is allowed
> before it, and ignored as far as the parser is concerned. Or is
> that not the case?

It is not. The text before su must still be Lojban; the text after
fa'o contains no such restriction.

-Robin



posts: 14214

On Sun, Nov 07, 2004 at 02:24:33PM -0800, Jorge Llamb?as wrote:
> --- Robin Lee Powell wrote:
differences between precedence and LTR versions
> > SI erases the preceding word unless it is a ZO. Y is ignored.
>
> This is probably the most different for the exceptionless rules.
>
> SI would NOT erase the previous word in the following cases: {zo
> si}, {zoi si ... si}, {da zei si}, {lo'u ... si ... le'u}, where
> the magic of SI is turned off by a preceding magic word.
>
> SI would erase MORE than a single word in the following cases: {zo
> a si}, {zoi zoi ... zoi si}, {da zei de si}, {lo'u ... le'u si},
> {a bu si}, {... sa si}, where SI erases the whole preceding magic
> word construct.

This is the stuff I'd really like people's opinions on. Anyone,
anyone?

> How is the selmaho rule supposed to work with respect to words
> grabbed by magic words? Do they count for SA?

No.

> For example:
>
> mi cusku zo broda sa bacru
>
> Does {bacru} replace {cusku} or {broda}?

{cusku}

> > ZOI cmavo use the following word as a delimiting word, no matter
> > what it is, execept Y (which is ignored); and SI, SA and SU
> > (which erase it).
>
> With exceptionless rules there would be no exceptions. (Y not
> counting as a word.)

My problem there is that backing out of a zoi is too damned hard.

zoi .y. si si si — that's the fastest way in LTR (assuming that the
average user would hesitate trying to figure out how to get out of
it). I think "zoi si" == nothing is important.

> > and FAhO (which makes no sense because the stream ends at the
> > FAhO).
>
> {da zei fa'o} is acceptable with the exceptionless rules because
> zei turns off the magic of fa'o.

Yeah, I think that's probably the right way to do it.

> > For words to its right, it does not affect SI, SA, and SU (which
> > erase it); BAhE and Y (which it skips); ZEI (which would lead to
> > grouping issues, and hence is an error to attempt); and FAhO
> > (which results in an error).
>
> With exceptionless rules, SI, SA, SU, BAhE, ZEI and FAhO are
> instead turned off by the preceding ZEI.

<nod> This makes it impossible to correct part of a ZEI, of course,
which could get painful.

-Robin


posts: 1912


> > 2) At least in Robin's parser implementation, FAhO has a true
> > magic word grammatical function besides its end-of-parsable-input
> > metagrammatical function. If I understand correctly, in Robin's
> > parser FAhO swallows also any preceding word that makes the
> > text ungrammatical.
>
> *NO*. This is *not* correct.
>
> PEG parsers, by default, output only those things they processed
> successfully. A grammatical error before fa'o causes that error and
> all that follows, *including* fa'o, to not be processed. Sort of.
> It's really more complicated than that.

OK. I misunderstood what you said on irc yesterday
about {<good stuff> <bad stuff> fa'o <bad stuff>}.
I understood you to say that that returned {<good stuff> fa'o},
but you must have said that it just returns {<good stuff>}, with
no {fa'o}.

I'd like to know what the official grammar of FAhO is, then.
Is it something like:

text = (rest of grammar) [FAhO anything]

?

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Wed, Nov 10, 2004 at 01:08:47PM -0800, Jorge Llamb?as wrote:
> I'd like to know what the official grammar of FAhO is, then. Is it
> something like:
>
> text = (rest of grammar) [FAhO anything]

Almost exactly like that, yes.

-Robin


posts: 1912


> > SI would NOT erase the previous word in the following cases: {zo
> > si}, {zoi si ... si}, {da zei si}, {lo'u ... si ... le'u}, where
> > the magic of SI is turned off by a preceding magic word.
> >
> > SI would erase MORE than a single word in the following cases: {zo
> > a si}, {zoi zoi ... zoi si}, {da zei de si}, {lo'u ... le'u si},
> > {a bu si}, {... sa si}, where SI erases the whole preceding magic
> > word construct.
>
> This is the stuff I'd really like people's opinions on. Anyone,
> anyone?

The last comment, about {... sa si} is not quite right. {sa} acts
first and since it won't find any SI looking back, it will erase
everything to the beginning of text. At that point SI has nothing
to erase, so it is either an error or a beginning of text exception
for SI.

> My problem there is that backing out of a zoi is too damned hard.
>
> zoi .y. si si si — that's the fastest way in LTR (assuming that the
> average user would hesitate trying to figure out how to get out of
> it). I think "zoi si" == nothing is important.

OTOH, backing out of {zoi zoi} is made easier:

zoi zoi .y. zoi si

instead of

zoi zoi .y. zoi si si si si

(BTW, what happens when "anything" is empty, do you still need
a {si} to erase it with your rules?)

So I don't think making the exception would be justified.


> > With exceptionless rules, SI, SA, SU, BAhE, ZEI and FAhO are
> > instead turned off by the preceding ZEI.
>
> <nod> This makes it impossible to correct part of a ZEI, of course,
> which could get painful.

Yes. It's impossible to correct parts of a very long normal
lujvo too. It's better not to use extremely long lujvo or zei-lujvo.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




Robin Lee Powell scripsit:

> This removes the primary function of fa'o, which is to allow
> arbitrary non-Lojban text afterwards.

In fact the primary purpose of fa'o is to separate multiple texts in a
single stream. It does *not* entail that the remainder is not Lojban,
merely that it is not part of the current text and should not be parsed

  • now*. That does not exclude parsing it in future.


--
All Norstrilians knew what laughter was: John Cowan
it was "pleasurable corrigible malfunction". http://www.reutershealth.com
--Cordwainer Smith, Norstrilia jcowan@reutershealth.com


posts: 14214

On Wed, Nov 10, 2004 at 04:33:25PM -0500, John Cowan wrote:
> Robin Lee Powell scripsit:
>
> > This removes the primary function of fa'o, which is to allow
> > arbitrary non-Lojban text afterwards.
>
> In fact the primary purpose of fa'o is to separate multiple texts
> in a single stream. It does *not* entail that the remainder is
> not Lojban, merely that it is not part of the current text and
> should not be parsed *now*. That does not exclude parsing it in
> future.

<blink>

The cmavo ``fa'o'' (of selma'o FAhO) is the usually omitted marker
for the end of a text; it can be used in computer interaction to
indicate the end of input or output, or for explicitly giving up
the floor during a discussion. It is outside the regular grammar,
and the machine parser takes it as an unconditional signal to stop
parsing unless it is quoted with ``zo or with ``lo'u ... le'u.
In particular, it is not used at the end of subordinate texts
quoted with ``lu ... li'u or parenthesized with ``to ... toi.

The Red Book seems to disagree with you. I note in particular
"outside the regular grammar" and "unconditional signal to stop".

I was under the impression that it was basically like typing ^D in
UNIX.

-Robin


posts: 14214

On Wed, Nov 10, 2004 at 01:29:24PM -0800, Jorge Llamb?as wrote:
> --- Robin Lee Powell wrote:
> > My problem there is that backing out of a zoi is too damned
> > hard.
> >
> > zoi .y. si si si — that's the fastest way in LTR (assuming that
> > the average user would hesitate trying to figure out how to get
> > out of it). I think "zoi si" == nothing is important.
>
> OTOH, backing out of {zoi zoi} is made easier:

I find it very unlikely that someone will say "zoi zoi" and *then*
realize they've made a mistake. Much more likely is "zoi broda",
having intended "zo broda", then realization, or realizing
immediately after zoi.

> (BTW, what happens when "anything" is empty, do you still need a
> {si} to erase it with your rules?)

Never thought about it. I think so, yeah.

> > > With exceptionless rules, SI, SA, SU, BAhE, ZEI and FAhO are
> > > instead turned off by the preceding ZEI.
> >
> > <nod> This makes it impossible to correct part of a ZEI, of
> > course, which could get painful.
>
> Yes. It's impossible to correct parts of a very long normal lujvo
> too. It's better not to use extremely long lujvo or zei-lujvo.

True.

I think I'm coming over to your side, but I'm worried about .y. bu.
I don't think we can just drop it out, much as we'd like to, but it
causes wierd problems like "zoi .y. bu...".

Would you accept "Before normal text processing, ".y. bu" is
handled, and is treated as a single word of selma'o BY during
regular LTR processing" as an exception? Do you see any problems
with this?

(My parser, of course, won't actually implement it as a seperate
pass. Not sure how I'll implement it at all, but it's certainly
possible. Probably a special case in BY itself.)

-Robin


posts: 1912


> I think I'm coming over to your side, but I'm worried about .y. bu.
> I don't think we can just drop it out, much as we'd like to, but it
> causes wierd problems like "zoi .y. bu...".
>
> Would you accept "Before normal text processing, ".y. bu" is
> handled, and is treated as a single word of selma'o BY during
> regular LTR processing" as an exception? Do you see any problems
> with this?

I don't see any problems with that at this point, other than
the conceptual ugliness.

> (My parser, of course, won't actually implement it as a seperate
> pass. Not sure how I'll implement it at all, but it's certainly
> possible. Probably a special case in BY itself.)

I tend to think of the morphology processing as separate from the
syntax processing. Your parser does everything together, so reading
the PEG grammar confuses me sometimes. I wonder if it's possible
to write the PEG grammar in such a way that the part that deals
with words only (i.e. not with individual characters) can stand
on its own.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

I've just put what I think will be my final version, barring strenuous "I would vote against this" sort of objections.

It was going to be closer to xorxes' version than John's, but that turned out not to be the case. The reason is very simple:

1. If you're going to allow the deletion of things piece by piece (zoi delimiters, for example) that'd better buy you something. John's version requires four si to delete a zoi but *provides no advantage to doing so*. You can't add anything on to the end of zoi once you've deleted the last delimiter; the *only* valid things that can follow are more si, anything else (except maybe another copy of the delimiter word) results in an error.

2. It is possible to allow the deletion of, say, zoi delimiters and the addition of more stuff to the zoi, but at the expense of requiring rules that end up with side effects like not being able to add things that start with the sound "si" to said zoi quote. That way madness lies, and as far as I can tell, everyone but Raphael would prefer not to go there.

So, there you have it. Barring any strenous objections as I said, xorxes wins again.

The main difference between what I've got and what xorxes has is the language; I like xorxes' ideas, in general, but I don't much like the way he phrases them. I stole the basic idea from John, but expanded on it a bit. Let me know what you think.

-Robin

posts: 14214

Just added some special SA handling, which I don't *think* introduces any bugs, but I need to play with it some more. As a side effect, one can now add on to a lo'u...le'u, although still not zoi.

I still don't know what fa'o means. Does fa'o si fo'a work, and if so, what's the point of having fa'o, exactly?

-Robin

posts: 1912


If {bu}, {zei} and {si} trump {le'u}, that's an exception
to left-to-right processing. The advantages I can see
would be:

(1) We can re-open the lo'u quote with {si} to keep adding words.
(2) We can sneak {le'u} inside the quote as {le'u bu} or {le'u zei}.

Any others?

(1) is not such a big gain because {joi lo'u} will get us
about the same thing, and saving two syllables for something
that will come up very rarely does not seem to justify making
an exception to the general left-to-right rule.

(2) Is it very important to be able to quote {le'u} with
{lo'u ... le'u}? I would prefer introducing a new cmavo,
say {lei'u}, whose only function is to represent {le'u}
within a LOhU-LEhU quote.

As for SA and BAhE, I think both ways are compatible with
left-to-right. I would prefer that SA does not see BAhE,
because it would seem that in many cases we might want to
emphasize the replacement word. We can still replace
{ba'e da} with {sa za'e da} as long as there was no
other intervening KOhA, for example.

In any case, SA is definitely not worth the trouble and
complications it causes. People are much more likely to
use {si} to make corrections anyway:

mi pilno le pinji lo nu ciska lo xatra
i pinji si pinsi

Trying to fix it with {sa} is too hard.

What happens with {... sa sa da}?

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 13, 2004 at 06:53:02PM -0800, Jorge Llamb?as wrote:
>
> If {bu}, {zei} and {si} trump {le'u}, that's an exception to
> left-to-right processing. The advantages I can see would be:
>
> (1) We can re-open the lo'u quote with {si} to keep adding words.
>
> (2) We can sneak {le'u} inside the quote as {le'u bu} or {le'u
> zei}.
>
> Any others?

Yes. In fact, it wasn't done for either of those reasons, it was
done so that SA LOhU and SA LEhU would work.

> (1) is not such a big gain because {joi lo'u} will get us about
> the same thing, and saving two syllables for something that will
> come up very rarely does not seem to justify making an exception
> to the general left-to-right rule.

That's a good point; hadn't thought of that.

> (2) Is it very important to be able to quote {le'u} with
> {lo'u ... le'u}?

Hell no; use zoi. That was just a side effect.

I'll go back to lo'u...le'u is all one big mess, of selma'o LOhU for
SA purposes, and reccomend the joi thing.

> As for SA and BAhE, I think both ways are compatible with
> left-to-right. I would prefer that SA does not see BAhE, because
> it would seem that in many cases we might want to emphasize the
> replacement word. We can still replace {ba'e da} with {sa za'e da}
> as long as there was no other intervening KOhA, for example.

  • Ewww*. You want SA to skip BAhE to the *right*? Gross.


> In any case, SA is definitely not worth the trouble and
> complications it causes. People are much more likely to use {si}
> to make corrections anyway:
>
> mi pilno le pinji lo nu ciska lo xatra
> i pinji si pinsi
>
> Trying to fix it with {sa} is too hard.

Oh, oh, that reminds me; we need a cmavo for s///. Or at least, I'd
very much like one. It would be of selma'o ZEI, and the semantics
are "pretend the last instance of the word on the left was actually
the word on the right". Oh, and now we can use zo with it and
everyhing. I love this idea, I'm glad I thought of it.

> What happens with {... sa sa da}?

As it says in the SA section, SA SA destroys back to the beginning
of text. So that would be just {da}.

-Robin


posts: 14214

> I'll go back to lo'u...le'u is all one big mess, of selma'o LOhU
> for SA purposes, and reccomend the joi thing.

This has the minor side effect that SA LEhU is *never* legal. Or,
to put it another way, always erases to the beginning of text.

That mean you can't continue a lo'u...le'u from far ahead. Maybe
that's not so minor.

-Robin


posts: 149

wikidiscuss@lojban.org scripsit:

> I still don't know what fa'o means. Does fa'o si fo'a work, and if so,
> what's the point of having fa'o, exactly?

The point is that nothing after fa'o is interpreted by this particular
event of parsing. What follows may be parsed by some other event.
If you were using a text-to-speech converter, you might convert fa'o

into
D/
Z, or you might convert it into ENTER. So fa'o cannot be

erased or otherwise messed with once it is recognized. It can be
protected by zo, lo'u...le'u, or zoi. fa'o zei broda won't work;
I don't care if broda zei fa'o works or not.

(Thought: should we have another member of ZEI with reverse-order
modified-modifier semantics?)

--
But that, he realized, was a foolish John Cowan
thought; as no one knew better than he cowan@ccil.org
that the Wall had no other side. http://www.ccil.org/~cowan
--Arthur C. Clarke, "The Wall of Darkness"


posts: 14214

> (Thought: should we have another member of ZEI with reverse-order
> modified-modifier semantics?)

Seems a waste of precious cmavo space to me.

-Robin


posts: 14214

On Sat, Nov 13, 2004 at 07:13:29PM -0800, Robin Lee Powell wrote:
> > I'll go back to lo'u...le'u is all one big mess, of selma'o LOhU
> > for SA purposes, and reccomend the joi thing.
>
> This has the minor side effect that SA LEhU is *never* legal. Or,
> to put it another way, always erases to the beginning of text.
>
> That mean you can't continue a lo'u...le'u from far ahead. Maybe
> that's not so minor.

I'm a moron.

"sa le'u" wouldn't work in any case; that would just re-close the
quote and continue on from there.

Will be fixed shortly.

-Robin


posts: 14214

On Sat, Nov 13, 2004 at 10:14:58PM -0500, John Cowan wrote:
> wikidiscuss@lojban.org scripsit:
>
> > I still don't know what fa'o means. Does fa'o si fo'a work, and
> > if so, what's the point of having fa'o, exactly?
>
> The point is that nothing after fa'o is interpreted by this
> particular event of parsing. What follows may be parsed by some
> other event.

Excellent. Fixed.

-Robin


posts: 1912


> *Ewww*. You want SA to skip BAhE to the *right*? Gross.

I don't exactly *want* that, it's just the least bad alternative
I can find. My preferences, in order, would be:

1- Eliminate SA from the language.
2- Change the meaning of SA to something simpler.
3- Use the emphasized/nonce word as the replacement word in
{SA BAhE <word>}.
4- Use BAhE as the replacement word.

> Oh, oh, that reminds me; we need a cmavo for s///. Or at least, I'd
> very much like one. It would be of selma'o ZEI, and the semantics
> are "pretend the last instance of the word on the left was actually
> the word on the right". Oh, and now we can use zo with it and
> everyhing. I love this idea, I'm glad I thought of it.

I wouldn't put it in selma'o ZEI though. That would only
allow single word replacements.

We already use {si} for single words, and now we can also use
{zo le si zo lo} and {lo'u da de di le'u si lo'u di de da le'u} for
replacing a word that is not grammatical by itself, or many
words.

If we want something more formal, perhaps we could somehow recycle
SA for this?

> > What happens with {... sa sa da}?
>
> As it says in the SA section, SA SA destroys back to the beginning
> of text. So that would be just {da}.

The section mentions SA SI and SA SU, not SA SA, but that's
what I would expect for SA SA too, yes.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 13, 2004 at 07:53:52PM -0800, Jorge Llamb?as wrote:
> > Oh, oh, that reminds me; we need a cmavo for s///. Or at least,
> > I'd very much like one. It would be of selma'o ZEI, and the
> > semantics are "pretend the last instance of the word on the left
> > was actually the word on the right". Oh, and now we can use zo
> > with it and everyhing. I love this idea, I'm glad I thought of
> > it.
>
> I wouldn't put it in selma'o ZEI though. That would only allow
> single word replacements.

Incorrect; it can take whole LOhU...LEhU now.

> We already use {si} for single words, and now we can also use {zo
> le si zo lo} and {lo'u da de di le'u si lo'u di de da le'u} for
> replacing a word that is not grammatical by itself, or many words.

Yes, but SI has a totally different semantic meaning.

> If we want something more formal, perhaps we could somehow recycle
> SA for this?

Unfortunately, the weight of usages like "le broda brode sa le brode
broda" is too strong for me to support radically altering SA.

> > > What happens with {... sa sa da}?
> >
> > As it says in the SA section, SA SA destroys back to the
> > beginning of text. So that would be just {da}.
>
> The section mentions SA SI and SA SU, not SA SA, but that's what I
> would expect for SA SA too, yes.

Oh, sorry. Fixed.

-Robin


posts: 1912



> > I wouldn't put it in selma'o ZEI though. That would only allow
> > single word replacements.
>
> Incorrect; it can take whole LOhU...LEhU now.

Yes, but the replacement can only be one word.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Sat, Nov 13, 2004 at 08:26:58PM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
>
> > > I wouldn't put it in selma'o ZEI though. That would only allow
> > > single word replacements.
> >
> > Incorrect; it can take whole LOhU...LEhU now.
>
> Yes, but the replacement can only be one word.

Huh? Oh, yeah. Good point.

A, then.

-Robin


On Saturday 13 November 2004 22:06, Robin Lee Powell wrote:
> On Sat, Nov 13, 2004 at 06:53:02PM -0800, Jorge Llamb?as wrote:
> > What happens with {... sa sa da}?
>
> As it says in the SA section, SA SA destroys back to the beginning
> of text. So that would be just {da}.

If you accidentally say {sa}, how do you erase it? I've seen {sa si} proposed
for this.

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 1912


Pierre wrote:
> If you accidentally say {sa}, how do you erase it?

With the word that came just before {sa}.

mu'o mi'e xorxes





__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 84

From the page:

> *ZO* binds with the following word. The combination is considered a
> single word of the pseudo selma'o /any-word/, except for SA matching
> purposes, where it retains the selma'o ZO.

Um, isn't the pseudo-selma'o really KOhA? Grammatically the quoted word
functions as a sumti. It's a single "any-word" wrt other Magic Word
processing, but its grammar is that of KOhA. Ditto uses of "any-string"
and such later in the document. Does this matter?

Also from the page:

da zei fa'o Lujvo of da and fa'o


Since you avoid xorxes' terminology of words "turning off" the magic of
other words, it isn't clear here that this {fa'o} doesn't terminate the
text-stream. Or that {da zei zei} doesn't suck up another following
word, but is just a brivla. You say that they are processed RTL, but I
think you need to be clearer about when these magic words lose their magic.

wikidiscuss@lojban.org, incarnated as Robin, wrote:

>Re: Magic Words
>Just added some special SA handling, which I don't *think* introduces any bugs, but I need to play with it some more. As a side effect, one can now add on to a lo'u...le'u, although still not zoi.
>
>
Probably okay. Adding onto zoi quotes is probably a Big Mistake
anyway. Adding onto lo'u/le'u quotes might be too, but if it can work,
maybe all right.

>I still don't know what fa'o means. Does fa'o si fo'a work, and if so, what's the point of having fa'o, exactly?
>
>
IMO, a non-quoted (non-disabled) fa'o is an end-of-file. {fa'o si
fo'a}, at best, begins a new text with {si fo'a}, which is allowed by
your rules. A helpful speaker would, of course, let you get away with it.

~mark



posts: 14214

On Sun, Nov 14, 2004 at 09:23:39PM -0500, Mark E. Shoulson wrote:
> From the page:
>
> >*ZO* binds with the following word. The combination is considered
> >a single word of the pseudo selma'o /any-word/, except for SA
> >matching purposes, where it retains the selma'o ZO.
>
> Um, isn't the pseudo-selma'o really KOhA?

That's not very pseudo, is it?

> Grammatically the quoted word functions as a sumti.

True, but so do many other things.

> It's a single "any-word" wrt other Magic Word processing, but its
> grammar is that of KOhA. Ditto uses of "any-string" and such later
> in the document. Does this matter?

I don't think so, but I will clarify it.

> Also from the page:
>
> da zei fa'o Lujvo of da and fa'o
>
>
> Since you avoid xorxes' terminology of words "turning off" the
> magic of other words, it isn't clear here that this {fa'o} doesn't
> terminate the text-stream. Or that {da zei zei} doesn't suck up
> another following word, but is just a brivla. You say that they
> are processed RTL, but I think you need to be clearer about when
> these magic words lose their magic.

Umm, zei binds with both words and turns the whole thing into a
brivla. zei gets there first (LTR processing). Not sure how that's
unclear. Can you give an idea of what you'd like to see added?

> wikidiscuss@lojban.org, incarnated as Robin, wrote:
>
> >Re: Magic Words Just added some special SA handling, which I
> >don't *think* introduces any bugs, but I need to play with it
> >some more. As a side effect, one can now add on to a
> >lo'u...le'u, although still not zoi.
>
> Probably okay.

I dropped it, though, after xorxes pointed out that "joi lo'u" only
costs one more syllable, and decreases complexity greatly.

> >I still don't know what fa'o means. Does fa'o si fo'a work, and
> >if so, what's the point of having fa'o, exactly?
>
> IMO, a non-quoted (non-disabled) fa'o is an end-of-file. {fa'o si
> fo'a}, at best, begins a new text with {si fo'a}, which is allowed
> by your rules. A helpful speaker would, of course, let you get
> away with it.

Agreed on all points.

-Robin


posts: 84

Jorge "Llambías" wrote:

>--- Robin Lee Powell wrote:
>
>
>>*Ewww*. You want SA to skip BAhE to the *right*? Gross.
>>
>>
>
>I don't exactly *want* that, it's just the least bad alternative
>I can find. My preferences, in order, would be:
>
>1- Eliminate SA from the language.
>2- Change the meaning of SA to something simpler.
>3- Use the emphasized/nonce word as the replacement word in
> {SA BAhE <word>}.
>4- Use BAhE as the replacement word.
>
>
I useta think SA did have a simpler meaning, something like "Erase
several words back; I'm not saying how many, though, you have to work
that out for yourself." Unspecified sort of like {ra}. Though
something like that sounds more like a UI usage than metasyntax.

>>Oh, oh, that reminds me; we need a cmavo for s///. Or at least, I'd
>>very much like one. It would be of selma'o ZEI, and the semantics
>>are "pretend the last instance of the word on the left was actually
>>the word on the right". Oh, and now we can use zo with it and
>>everyhing. I love this idea, I'm glad I thought of it.
>>
>>
>
>I wouldn't put it in selma'o ZEI though. That would only
>allow single word replacements.
>
>
It would also make the result be a brivla hanging out in the middle of
nowhere, whereas what you want is something like a UI or other freemod.
It doesn't mean anything much right here, but affects meaning earlier
on... You might be able to do something with SEI, but it's a mess
waiting to happen.

~mark


According to the Magic Words page, {ybu} "is considered a single word of
selma'o BU" (I suppose you mean BY) and this is done first. What does {zo y
bu} turn into, as compared with {zo a bu}?

phma
--
li ze te'a ci vu'u ci bi'e te'a mu du
li ci su'i ze te'a mu bi'e vu'u ci


posts: 1912


> According to the Magic Words page, {ybu} "is considered a single word of
> selma'o BU" (I suppose you mean BY) and this is done first. What does {zo y
> bu} turn into, as compared with {zo a bu}?

{zo y bu} = the word 'ybu'
{zo a bu} = the lerfu 'a'

This is the ugly consequence of introducing exceptions. We could
introduce further exceptions and preprocess {a bu}, {e bu}, {i bu},
{o bu} and {u bu} as well, but one exception always leads to another
and the result is a mess.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Mon, Nov 15, 2004 at 08:33:33AM -0800, Jorge Llamb?as wrote:
>
> --- Pierre Abbat wrote:
> > According to the Magic Words page, {ybu} "is considered a single
> > word of selma'o BU" (I suppose you mean BY) and this is done
> > first. What does {zo y bu} turn into, as compared with {zo a
> > bu}?
>
> {zo y bu} = the word 'ybu'
>
> {zo a bu} = the lerfu 'a'
>
> This is the ugly consequence of introducing exceptions.

Yep.

> We could introduce further exceptions and preprocess {a bu}, {e
> bu}, {i bu}, {o bu} and {u bu} as well, but one exception always
> leads to another and the result is a mess.

I agree. Let's not.

If I thought I could get away with dropping "y bu" entirely, I
would.

-Robin


posts: 1912


> If I thought I could get away with dropping "y bu" entirely, I
> would.

In such case, we could use {yb bu} for the lerfu in question.

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




posts: 14214

On Mon, Nov 15, 2004 at 11:01:24AM -0800, Jorge Llamb?as wrote:
>
> --- Robin Lee Powell wrote:
> > If I thought I could get away with dropping "y bu" entirely, I
> > would.
>
> In such case, we could use {yb bu} for the lerfu in question.

  • If*, xorxes, *if*. Subjunctive. {da'i}.


Unless you can get me, say, half a dozen BPFK members who all say
they'd support this, it ain't happening.

-Robin


posts: 1912


> On Mon, Nov 15, 2004 at 11:01:24AM -0800, Jorge Llamb?as wrote:
> > --- Robin Lee Powell wrote:
> > > If I thought I could get away with dropping "y bu" entirely, I
> > > would.
> >
> > In such case, we could use {yb bu} for the lerfu in question.
>
> *If*, xorxes, *if*. Subjunctive. {da'i}.

That's why I used the conditional "could". In such case, i.e. if
you thought you could get away with it and therefore you would
drop it, then we could use {yb bu} for the lerfu.

> Unless you can get me, say, half a dozen BPFK members who all say
> they'd support this, it ain't happening.

We already have two, we only need four more. Anyone else?

mu'o mi'e xorxes




__
Do you Yahoo!?
Check out the new Yahoo! Front Page.
www.yahoo.com




Robin Lee Powell scripsit:

> > In such case, we could use {yb bu} for the lerfu in question.
>
> *If*, xorxes, *if*. Subjunctive. {da'i}.
>
> Unless you can get me, say, half a dozen BPFK members who all say
> they'd support this, it ain't happening.

I'd support it. It's a wart, but *everything* in this area is a wart.

--
John Cowan http://www.ccil.org/~cowan <jcowan@reutershealth.com>
You tollerday donsk? N. You tolkatiff scowegian? Nn.
You spigotty anglease? Nnn. You phonio saxo? Nnnn.
Clear all so! `Tis a Jute.... (Finnegans Wake 16.5)


posts: 14214

On Mon, Nov 15, 2004 at 10:07:09PM -0500, John Cowan wrote:
> Robin Lee Powell scripsit:
>
> > > In such case, we could use {yb bu} for the lerfu in question.
> >
> > *If*, xorxes, *if*. Subjunctive. {da'i}.
> >
> > Unless you can get me, say, half a dozen BPFK members who all
> > say they'd support this, it ain't happening.
>
> I'd support it. It's a wart, but *everything* in this area is a
> wart.

You'd support dropping {y bu} from the language? I'm quite
surprised.

-Robin

--
http://www.digitalkingdom.org/~rlpowell/ *** http://www.lojban.org/
Reason #237 To Learn Lojban: "Homonyms: Their Grate!"
Proud Supporter of the Singularity Institute - http://singinst.org/


posts: 84

I don't remember seeing this handled in the discussions (which I've read, actually), and it isn't answered in the page itself. Can SA see inside a LU/LI'U quote? We know it can't see into a LO'U/LE'U quote to find its word to erase, but what about LU/LI'U? Both "yes" and "no" answers are probably defensible, but we need to decide on *one* and make it Official.

Should we be clearer regarding the whole "Left to Right" business? It makes sense to us now that we've gone through this whole debate, but we are writing for the ages, so we should be explicit as to what sorts of conflicts arise and how they are resolved.

~mark

posts: 14214

On Sat, Jan 01, 2005 at 06:00:59PM -0800, wikidiscuss@lojban.org wrote:
> I don't remember seeing this handled in the discussions (which
> I've read, actually), and it isn't answered in the page itself.
> Can SA see inside a LU/LI'U quote?

Yes, absolutely. There's no reason *not* to, that I can see.

> Should we be clearer regarding the whole "Left to Right" business?
> It makes sense to us now that we've gone through this whole
> debate, but we are writing for the ages, so we should be explicit
> as to what sorts of conflicts arise and how they are resolved.

Do you have suggestions as to how to be clearer?

-Robin


wikidiscuss@lojban.org scripsit:
> Re: Magic Words
> I don't remember seeing this handled in the discussions (which I've
> read, actually), and it isn't answered in the page itself. Can SA see
> inside a LU/LI'U quote? We know it can't see into a LO'U/LE'U quote
> to find its word to erase, but what about LU/LI'U? Both "yes" and
> "no" answers are probably defensible, but we need to decide on *one*
> and make it Official.

I think the answer is clearly "yes". A lu...li'u quotation is forced to be
grammatical because it's in effect part of the outside text.

--
"They tried to pierce your heart John Cowan
with a Morgul-knife that remains in the http://www.ccil.org/~cowan
wound. If they had succeeded, you would http://www.reutershealth.com
become a wraith under the domination of the Dark Lord." --Gandalf


Re: Magic Words
Why, exactly, is BAhE + BU invalid? The Red Book (C19S16 as usual) seems quite clear on this point:

``bu makes the preceding word into a lerfu word, except for ``zo, ``si, ``sa, ``su, ``lo'u, ZOI cmavo, ``fa'o, ``zei, BAhE cmavo, and ``bu. Multiple ``bu cmavo may be used in succession.

-Robin



My Friend SU
Does SU go back to the last NIhO, LU, TUhE, or TO (as grammar.300 claims) or the beginning of input (as the Red Book claims)?

-Robin



Re: Magic Words
grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU clauses. I'm pretty sure this is just a YACC side effect; it certainly seems amazingly ridiculous to me. If anyone disagrees, let me know.

I suppose I can imagine something like "mi zoi zoi quietude zoi zei cinmo", but I don't know if it's worthwhile building such a bizarre construction into the language. Maybe.

ZO-clause+ZEI and LOhU...LEhU+ZEI are just dumb as hell, though.

-Robin



Re: Magic Words
The red book claims that UI cmavo can't mark BU. What's up with that?

Specifically, it says:

UI and CAI cmavo mark the previous word, except for ``zo, ``si, ``sa, ``su, ``lo'u, ZOI, ``fa'o, ``zei, BAhE cmavo, and ``bu. Multiple UI cmavo may be used in succession. A following ``nai'' is made part of the UI.

-Robin



Re: Magic Words
What do we do about SI, SA and ZOI?

Option 1:

  • SI is an allowed ZOI delimiter.
  • SI after ZOI eats through delimiters and text seperately, so four SI in a row are required to erase a ZOI.
  • Fastest way to erase a ZOI clause is to use SA + something that occured before the ZOI; SA+ZOI doesn't help at all; it just puts you back in the ZOI. Since SI can be a ZOI delimiter, SA+ZOI+SI is just a ZOI clause ready for non-Lojban text.


Option 2:

  • SI is an allowed ZOI delimiter.
  • SI after ZOI eats the entire thing, ZOI and all.
  • SA+ZOI puts you back in the ZOI. Since SI can be a ZOI delimiter, SA+ZOI+SI is just a ZOI clause ready for non-Lojban text.


Option 3:

  • SI is NOT an allowed ZOI delimiter.
  • SI after ZOI eats through delimiters and text seperately, so four SI in a row are required to erase a ZOI.
  • SA+ZOI puts you back in the ZOI. Since SI can NOT be a ZOI delimiter, SA+ZOI+SI is the fastest way to delete the whole thing.
  • This would mean that the only case where WORD + SA + same category as WORD + SI doesn't == nothing would be with ZO.


I have a strong preference for option three; it gives the most user friendliness, IMO. What do you all think?

-Robin



Re: Magic Words

Definition: A magic word is any cmavo of selmaho ZO, ZOI, LOhU, LEhU, ZEI, BU, SI, SA, SU, BAhE or FAhO
Definition: Y is not a word.

The exeptionless rules for magic words would be as follows:

ZO: Unless it has been turned off by a preceding magic word, it quotes the following word and, if it is a magic word, turns it off. If there is no following word, it gives an error.

ZOI: Unless it has been turned off by a preceding magic word, it turns the following word into a delimiter and, if it is a magic word, turns it off.
Everything that follows is non-lojban-text until the delimiter is found again (turned off if it's a magic word). If there is no following word, or if the following word does not reappear to close the quote, it gives an error.

LOhU: Unless it has been turned off by a preceding magic word, it quotes each following word until LEhU and, if it is a magic word, turns it off. If LEhU does not appear, it gives an error.

LEhU: Unless it has been turned off by a preceding magic word, it closes a LOhU quote. If it's not turned off and there is no opened LOhU it gives an error.

ZEI: Unless it has been turned off by a preceding magic word, it takes the previous word or magic word output, and joins it with the following word to make a tanru-unit. If the following word is a magic word, it turns it off. If there is no previous stuff, or if there is no following word, then it gives an error.

BU: Unless it has been turned off by a preceding magic word, it takes the previous word or magic word output, and turns it into a lerfu.

SI: Unless it has been turned off by a preceding magic word, it takes the previous word or magic word output and anihilates it. (i.e. now the previous stuff is the stuff before the one erased). If there is no previous stuff, it gives an error.

SA: Unless it has been turned off by a preceding magic word, it deletes everything back up to and including the first word found of the selmaho of the word that follows (stuff processed by preceding magic words in general does not keep its original selmaho) or else if no such word is found, up to the beginning of text. If no word follows, it gives an error.

SU: Unless it has been turned off by a preceding magic word, it deletes everything back up to the beginning of text. It never gives error.

BAhE: Unless it has been turned off by a preceding magic word, it modifies the word that follows, and becomes invisible (i.e. it is not part of the previous stuff for following magic words). If what follows is a magic word, it does not turn it off. If no word follows, it gives an error.

FAhO: Unless it has been turned off by a preceding magic word, it terminates the text. If the utterance so far is not a valid text, it gives an error.

Those rules are exeptionless, and they don't require to define any special order of precedence. The order is simply first come, first served. In some cases we may want, for whatever reason, to deviate from the exceptionless rules. For example, if we wanted {si} to erase {bu}, or {zoi}, or {zei}, or even {zo} for that matter. But the price of any exception is complication of the rules and opening a can of worms, because usually exceptions require more exceptions to handle special cases. I propose we start from the exceptionless rules and analyse carefully any proposed exception to them.

mu'o mi'e xorxes




Re: Magic Words

This are the answers to the Outstanding Questions, according to the exceptionless rules:

> Is nested lo'u...le'u allowed?

No, {lo'u} turns off every magic word that follows, including other {lo'u}s, up to the first {le'u} found.

> Not really, but lo'u has no effect and we allow "zo le'u".

The exceptionless rules do not allow {zo le'u} within the quote because {zo} is turned off and {le'u} then terminates it.

> Does zoi function in lo'u...le'u?
> No. If you need to quote a broken zoi, use another zoi.

Right.

> What does BAhE+BU do?
> Just empasizes the BU.

Right.

> The red book claims that UI cmavo can't mark BU. What's up with that?
> I have no idea, but I'm ignoring it. UI can mark anything that doesn't grab it.

Good. I have no idea what's up with that either.

> What do we do about ZOI, SI and SA interactions? Is SI allowed
> as a ZOI delimiter? Does SI after a ZOI clause erase the whole
> thing?

The exceptionless rules would accept SI and SA as delimiters.
SI after a ZOI-clause would erase the whole thing.

> SI, SA and SU are not allowed ZOI delimiters.
> 4 SI for ZOI erasure from outside.
> SA+ZOI+SI works. ZOI+SI == nothing.

Those would constitute exceptions to the exceptionless rules.

> Does SU go back to the last NIhO, LU, TUhE, or TO
> (as grammar.300 claims) or the beginning of input
> (as the Red Book claims)?
> The latter; SA can be used for the other things, and if SU works
> the former way there is no way to unequivocably erase to
> the beginning of input.

Good. Either way could work with the exceptionless rules, though.

> grammer.300 claims that ZEI works on ZOI, ZO and LOhU...LEhU
> clauses.

That agrees with the exceptionless rules. To clarify: ZEI would take a ZOI, ZO or LOhU-LEhU clause to its left, but just ZO, ZOI, LOhU or LEhU to its right. This is because ZEI cannot turn the magic off of those words once they have used it.

> I'm pretty sure this is just a YACC side effect; it certainly seems
>amazingly ridiculous to me. If anyone disagrees, let me know.
> It doesn't do any harm, and allows certain useful things (like ZEI
> lujvo with the various words that ZEI can't bind to in them).

And most importantly: It simplifies the rules.

mu'o mi'e xorxes




Re: Magic Words

I inadvertently left FAhO out when doing the formal rules
for the magic words in "exceptionless rules" mode.
Would the following do what I want for FAhO?

text = (rest of the grammar) faho-stop

faho-stop = FAhO / any-word faho-stop

What I would want is for FAhO to absorb any extra word
as needed to leave a grammatical text behind. Is that
how FAhO is supposed to work?

mu'o mi'e xorxes




Re: Magic Words

Some more comments.

> ZO quotes the following word, no matter what it is, except Y.
> Words quoted with ZO lose their grammatical functions.

This doesn't mention {zo zo da} and {zoi zo da ........ zo de}.

If we can't assume left-to-right precedence, and {zo}
wins over everything else, then in those cases {da} and {de}
would be quoted by {zo}. Also the difference between
{zo da si de} and {zo da si si} should be mentioned here.

> FAhO terminates the word or text stream unequivocably,
> unless quoted with ZO.

Even within {lo'u ... le'u}?

> LOhU quotes all following Lojban words up to a LEhU (but
> not a ZO+LEhU; this is to allow nested LOhU...LEhU quotes
> inside a LOhU...LEhU, so you can talk about mistakes you
> made that include a previous error quote). Except for the
> ZO+LEhU case (which should be read as simply LEhU inside
> a LOhU...LEhU quote) all Lojban words within a LOhU...LEhU
> quote are read without any grammar and have no
> grammatical effects.

Even FAhO? If FAhO is active within LOhU-LEhU, what about
ZO+FAhO?

> LEhU is ungrammatical except at the end of a LOhU quotation
> and after ZO.

Is {le'u si} ungrammatical?

> ZOI cmavo use the following word as a delimiting word, no
> matter what it is, execept Y (which is ignored); and SI, SA
> and SU (which erase it).

Can it use ZO, FAhO, LEhU?

> BU makes the preceding word into a lerfu word, except for ZO,
> ZEI and LOhU (which quote it), ZOI (which uses it as a delimiter);
> LEhU (which would result in bizarrely re-opening the LOhU...LEhU,
> and hence is an error); SI, SA and SU (it affects whatever is to the
> left after the erasing is done; if nothing is left an error results);
> BAhE (which it skips); BU (which would lead to grouping issues,
> and hence is an error to attempt); and FAhO (after which anything,
> including BU, is ignored).

What about {zo da bu} and {da zei de bu}?

mu'o mi'e xorxes




Re: Magic Words
A simple question:

How many of you think that the rules as I currently have them written would be very hard for you to learn?

For those of you who have also read xorxes' stuff, do you find it easier?

-Robin



On Tue, 9 Nov 2004, Robin Lee Powell wrote:

>> BU makes the preceding word into a lerfu word, except for ZO, ...
>> BU (which would lead to grouping issues, and hence is an error to
>> attempt); ... Multiple BU may be used in succession, in which case
>> a new letteral is formed for each additional BU (i.e. "broda bu"
>> is a different letteral from "broda bu bu"). However, "bu bu" by
>> itself is illegal.
>>
>> That seems to contradict itself about bu bu. Should the "BU
>> (which would lead to grouping issues, and hence is an error to
>> attempt);" part be stricken?
>
> "bu bu" by itself is an error. "da bu bu" is not.

The problem is the part that says that BU works for everything except for "BU
(which would lead to grouping issues, and hence is an error to attempt)".
But then you specifically go on to say that multiple BUs are fine, which
seems to mean that it's not an error to attempt. I think that sort of got
copied from the description for zei, and then not updated as the actual cases
got resolved later on.
--
Adam Lopresto
http://cec.wustl.edu/~adam/

dei se du'u nei

Re: Magic Words
I've just put what I think will be my final version, barring strenuous "I would vote against this" sort of objections.

It was going to be closer to xorxes' version than John's, but that turned out not to be the case. The reason is very simple:

1. If you're going to allow the deletion of things piece by piece (zoi delimiters, for example) that'd better buy you something. John's version requires four si to delete a zoi but *provides no advantage to doing so*. You can't add anything on to the end of zoi once you've deleted the last delimiter; the *only* valid things that can follow are more si, anything else (except maybe another copy of the delimiter word) results in an error.

2. It is possible to allow the deletion of, say, zoi delimiters and the addition of more stuff to the zoi, but at the expense of requiring rules that end up with side effects like not being able to add things that start with the sound "si" to said zoi quote. That way madness lies, and as far as I can tell, everyone but Raphael would prefer not to go there.

So, there you have it. Barring any strenous objections as I said, xorxes wins again.

The main difference between what I've got and what xorxes has is the language; I like xorxes' ideas, in general, but I don't much like the way he phrases them. I stole the basic idea from John, but expanded on it a bit. Let me know what you think.

-Robin



Re: Magic Words
Just added some special SA handling, which I don't *think* introduces any bugs, but I need to play with it some more. As a side effect, one can now add on to a lo'u...le'u, although still not zoi.

I still don't know what fa'o means. Does fa'o si fo'a work, and if so, what's the point of having fa'o, exactly?

-Robin



Re: Magic Words
I don't remember seeing this handled in the discussions (which I've read, actually), and it isn't answered in the page itself. Can SA see inside a LU/LI'U quote? We know it can't see into a LO'U/LE'U quote to find its word to erase, but what about LU/LI'U? Both "yes" and "no" answers are probably defensible, but we need to decide on *one* and make it Official.

Should we be clearer regarding the whole "Left to Right" business? It makes sense to us now that we've gone through this whole debate, but we are writing for the ages, so we should be explicit as to what sorts of conflicts arise and how they are resolved.

~mark