History: BPFK Section: ZOI

This page resulted from a discussion about the way ZOI is handled in camxes and the specification for it in the CLL:

http://groups.google.com/group/lojban/browse_frm/thread/aa27ff25e6dd5201

zoi, as described in the CLL:

http://dag.github.com/cll/19/10/

states, about delimiters:

"...and which is not found in the written text or spoken phoneme stream."

"Within written text, the Lojban written word used as a delimiting word may not appear..."

It then goes on to provide an example that is claimed to be ungrammatical:

mi djuno fi le valsi po'u zoi gy. gyrations .gy.

One could infer, given the example provided, that "gy" could not appear *at all* in the quoted text. The official parser (grammar.300) does not behave this way, however, as it breaks up the words into tokens and fails to detect that the delimiter might be a substring of a quoted word.

camxes also allows the form above as quoted by the CLL. It does not allow the following:

mi djuno fi le valsi po'u zoi gy. gyrate .gy.

because "gyrate" are three separate Lojban words, and camxes parses the quoted text as if it were Lojban, stopping after matching the first parsed word against the terminator.

Exactly how a parser should parse text between zoi delimiters needs to be decided. There are three extant proposals:

1. Consider the PEG grammar is it is written now to be correct and update the CLL to more accurately describe how zoi works.
2. replace the rule for zoi-word to match non-Lojban words (strings of non-whitespace) rather than lojban words, so 'gyrate' won't be divided into three words. The CLL will need to be updated for this behavior too.
3. Replace the PEG grammar with something that reads the stuff between the ZOI delimiter a character at a time. Either requiring a pause before the final delimiter or not. This is consistent with the CLL.