On 2012/09/02 11:52:46, dak wrote:
It looks like some _severe_ doctoring around with regard to notenames
was done
to make regtests pass without an actual understanding of the failure
modes,
introducing some half-baked in-between modes that don't have a purpose
apart
from papering over the fact that this patch causes the lexer to be in
the wrong
mode due to parser lookahead at several points of time.
My desire is for just one in-between mode, which I want for scanning top-level. That desired mode is to have pitch-names loaded, but without the scanner accepting music with durations. I want to talk about it while it is un-baked. The difficulty in putting the lexer in the correct mode, after music functions with variable argument lists, is related to the mailing list discussion on "GLISS". This is the version that allows digits at the *end* of identifiers, which might not be wise. See the tracker <http://code.google.com/p/lilypond/issues/detail?id=1670> The regression test changes demonstrate its conflict with possibly-useful syntax; I was surprised how few changes were needed. I have been waiting with this for over a year, so of course I will not try to push anything forward until this "GLISS" dicussion is done. http://codereview.appspot.com/6493072/ _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
