What follows is a proposal= , not a description of the current state of affairs. i believe that it is b= ackwards compatible with most uses, although possibly not all edge cases.

A "text" is everything produced by the curren= t rule named "text" in the formal grammar. This is a single chunk= of input, and can be fed to a parser by itself. It has an elidible termina= tor ({fa'u}) to explicitly indicate where it ends. It also has a new in= itiator to explicitly mark the beginning of parseable text. I'll use {f= a'ai}, mnemonic {cfari}, but I don't care what word it is. In the e= nd, it may end up analyzed out, but for now it's helpful to have it. {f= a'ai} is of its own selma'o.

A "conversation" is a sequence of zero or mor= e texts, optionally with non-lojban garbage between them.=A0

There is also a new continuation word (tentatively {di'ai}, = since that's what was used earlier) of its own selma'o. if it occur= s at the beginning of a text (after optional {fa'ai}), we treat the tex= t as a continuation of the preceding text (ignoring everything between them= ). =A0

"conversation" is the new top-level rule in t= he parser. it gracefully degrades to just a single text.

So given an exchange in IRC

A: do klama ma<= /div>

B: lo zarci

We can either parse each line com= pletely independently, or we can make our logs in a parseable format thus:<= /div>

A: fa'ai do klama ma fa'o

B: fa&= #39;ai lo zarci fa'o

That entire block is a single lojban conversation, and = can be parsed.

Potential issues:

The conversation rule should ignore leading garbage before the firs= t {fa'ai}, but {fa'ai} should be optional. This may mean that we mi= sparse text or parse what we shouldn't, only to throw it away. I don= 9;t know whether that's workable or not, but it seems similar to issues= {su} has.

This view holds with the idea that continuations are al= ways explicit. If you want a new text to be treated as part of an old text,= start it with {di'ai}. If you want to be explicit that it is *not* par= t of the preceding text, {fa'ai} will do that. Otherwise, it will impli= citly be a new text, but if context is strong enough people may ignore that= for practical reasons.

How these work with other Magic Words is left as an exe= rcise for someone else, for now.=A0

--
You received this message because you are subscribed to the Google Groups "= BPFK" group.
To post to this group, send email to bpfk-list@googlegroups.com.
To unsubscribe from this group, send email to bpfk-list+unsubscribe@googleg= roups.com.
For more options, visit this group at http://groups.google.com/group/bpfk-l= ist?hl=3Den.
--0016364d31b79666b604926eb1d4--