Suppressed a z0 custos separated from a line break (z) by only bars/clefs.

[henryso]

Fixes #1190.

Some tests (that happen to employ this sequence of figures) change, and a test was added to exercise all the variants in the issue post.

Please review and merge if satisfactory.

[eroux]

Well, I think the most important part is that now things like (c4z) will work properly.

Edit: sorry, I didn't realize you din't implement it yet!

[eroux]

Well, the changes look OK to me, but I don't think they really make sense without the changes that make (c4z) work properly (meaning: just changing clef, not prinint it)

[henryso]

Ok, I'm confused. How is (c4z) supposed to change?

[eroux]

Hmmm, I realize I wasn't clear in my initial comment, for instance in #1190 (comment) what I call a "clef change" is not really typing the clef, it's just instructing GregorioTeX to change the clef at the beginning of lines... Basically, the idea is that when a z is present in the area of a clef change, we don't make a discretionary with the different case (line breaking here or not), we directly typeset the relevant case. For instance the output of (z0::c4z) in GregorioTeX would be the same as the one (::z) plus the instruction to change the clef at the beginning of lines. Do you see what I mean?

[henryso]

Ok, if I understand correctly, a clef immediately before a line break in GABC should be typeset as an instruction to change the "line clef" starting with the line after the line break. Is that correct?

[henryso]

This gets tricky positioning the custos in the localrightbox before such a clef change. Any ideas would be appreciated.

[eroux]

It's correct yes, and the discretionary mechanism should not be used. For the custos, I think it's computed in the exact same way as if it was a z0, what would be the difference? If what you mean is that there is no macro to tell GregorioTeX to use an end of line custos without typesetting it directly I can add one

[henryso]

I need help here. I made the changes, and the executable emits \GreNextCustos and \GreSetLinesClef instead of a "real" clef when the clef is at the end of a line. I think it's emitting the correct thing, but the localleftbox does not seem to be updated. The localrightbox is correctly placing the custos for the clef change. Any ideas what I'm doing wrong?

[eroux]

Thanks a lot, I'll take a look tomorrow

[eroux]

Well, these days seem very busy, I'm not sure I'll be able to test, I'll tell you when I can (maybe not before next week). @rpspringuel maybe you would have time to take a look and debug the TeX side?

[eroux]

Trying the pull request I'm getting:

gabc/gabc-score-determination-l.c: In function ‘yy_get_next_buffer’:
gabc/gabc-score-determination.l:73:23: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
         for (n = 0; n < max_size \
                       ^
gabc/gabc-score-determination-l.c:1941:3: note: in expansion of macro ‘YY_INPUT’
   YY_INPUT( (&YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[number_to_move]),
   ^~~~~~~~
gabc/gabc-score-determination-l.c: In function ‘gabc_score_determination__scan_bytes’:
gabc/gabc-score-determination-l.c:2453:17: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
  for ( i = 0; i < _yybytes_len; ++i )
                 ^
mv -f gabc/.deps/gregorio-gabc-score-determination-l.Tpo gabc/.deps/gregorio-gabc-score-determination-l.Po
gcc -DHAVE_CONFIG_H -DBRANCH_VERSION='"henryso-fix-1190-9bb3fd0-3898"' -I.  -I../src -I../src/gabc  -I../src/dump  -I../src/gregoriotex -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2   -g -O2 -std=gnu89 -pedantic -fstack-protector-strong -fPIE -Wformat=2 -Werror=format-security -Wstrict-prototypes -Wdeclaration-after-statement -Wall -Wextra -MT gabc/gregorio-gabc-notes-determination-l.o -MD -MP -MF gabc/.deps/gregorio-gabc-notes-determination-l.Tpo -c -o gabc/gregorio-gabc-notes-determination-l.o `test -f 'gabc/gabc-notes-determination-l.c' || echo './'`gabc/gabc-notes-determination-l.c
gabc/gabc-notes-determination-l.c: In function ‘yy_get_next_buffer’:
gabc/gabc-notes-determination-l.c:11599:18: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
   for ( n = 0; n < max_size && \
                  ^
gabc/gabc-notes-determination-l.c:13259:3: note: in expansion of macro ‘YY_INPUT’
   YY_INPUT( (&YY_CURRENT_BUFFER_LVALUE->yy_ch_buf[number_to_move]),
   ^~~~~~~~
gabc/gabc-notes-determination-l.c: In function ‘gabc_notes_determination__scan_bytes’:
gabc/gabc-notes-determination-l.c:13763:17: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
  for ( i = 0; i < _yybytes_len; ++i )

(with gcc6 on debian/sid)

[eroux]

Ok, nailed it: \gre@save@clef also needs to be called... I don't have time anymore tonight, but I think the best would be to call it from \GreSetLinesClef directly, I'm not really sure why it's not done that way....

[henryso]

Thanks. I made the change, reviewed the tests and accepted those changes as well.

This is now ready for review and merge if satisfactory.

[eroux]

Well, everything looks ok to me, maybe we could just add a test with z- and Z-?

[henryso]

Done. Please review.