Re: [lojban] Re: zoizoi

[Pierre Abbat <phma@webjockey.net>]

On Wednesday 08 January 2003 19:02, Pierre Abbat wrote:
> I am currently starting the brivla-end-breaking routine, which is simpler
> than the brivla-start-breaking routine. This case, or actually a slight
> variant of it, tells me that the cmavo preceding a brivla have to be broken
> off before the brivla is broken from what follows it. Consider
> /zoiFLAluKAVbu.blableblibloblu.FLAluKAVbu./. If the part after the brivla
> is broken off before or at the same time as the part before it, this will
> be /zoi/ /FLAlu/ /KAVbu/ /blableblibloblu/ /FLAluKAVbu/ when the {zoi} is
> detected, and the lexer will erroneously call {kavbu blableblibloblu} the
> quote.

On further thought, it's going to be difficult to catch all errors of this 
sort. Consider /zoikybuladjAn/. This is lexed as follows:
It ends in a consonant. Look backward for a cmegadri and break before and 
after:
/zoikybu/ /la/ /djAn/
Look backward for [yC] or [C*V]. The first it finds is /yb/, so break there:
/zoiky/ /bu/ /la/ /djAn/
The first part consists entirely of cmavo. Split it:
/zoi/ /ky/ /bu/ /la/ /djAn/
At this point it recognizes {zoi} and takes {ky} to be the delimiter, not 
noticing that it was originally not followed by a pause.

phma