Re: [lojban] Re: cmegadri valfendi preti

["Lionel Vidal" <nessus@free.fr>]

Pierre Abbat:
> Forward parsing does not work well for Lojban because it does not have the
> prefix property. jbofi'e uses forward parsing; the result is an unwieldy
> state table and difficult-to-fix bugs (e.g. it accepts {ckankua} as a
> fu'ivla
> but does not accept {packankua}, even though it breaks cmavo off lujvo and
> type-3 fu'ivla).

I strongly disagree, as forward parsing is actually what you do as human
being when verbally communicating (as opposed to reading when
you have the time to try the current like  algorithm). Not that I have that
kind of experience myself, which I concede could be a valid argument
to consider what I said as unfounded :-)
I have not read the code of jbofi'e, but the quality and ease of maintenance
of a given implementation does not prove anything on the quality of
the underlying algorithm, not even considering the non-unicity of such
algorithms.

-- Lionel