bleeding-eyes diff (zapping perl regexps from sed invocations) from
jirib, who also tested the complete diff.
According to jirib, it's still very slow (ten minutes for two pages),
but at least it works.
Let it build with png-1.5 (from naddy@, who surely went blind on it).
While here, patch away some weird attempts to get something meaningful
out of code like this (yes, I show it to you all, you have to share
the pain):
if ( timezone==0 )
fprintf( pi->out, "Z)\n" );
else
fprintf( pi->out, "%+02d')\n", (int) timezone/3600 ); /* doesn't handle half-hour zones */
That's the address of timezone(3), divided by 3600 m(
ok naddy@
This still uses the downloaded documentation instead of building
it, because it's no fun do build the documentation on low-memory
and/or slow systems like arm (it ran for a couple of days on my
armish, and then it failed with a segfault, which i then couldn't
reproduce for the specific file the segfaults happend on).
It also doesn't depend on the texinfo-4.13 port I sent to ports@
(but not yet committed) recently.
Running it on some of my (rather oldish, i.e. written for lilypond-2.10
and older) files produces mixed results -- simple stuff still works,
some stuff reports failure (yet creates correct output, which is
strange). Complicated music like my (never finished) work on the
Busoni transcription of BWV-1004/Partita II doesn't work at all
(but iirc, it didn't with lilypond-2.12, either).
make regress currently fails with this version, because it would
need the texinfo port and a new test-baseline, which I didn't yet
create.
Some work to build the documentation for real (possibly shielded by a
pseudo flavor) will follow.
Drop gstoraster (formerly known as ps2raster); it is replaced
by the new gstoraster package.
Also Drop pstopxl, it's currently broken (at least since gs-9.00).
If anyone needs it, please let me know (or patch/repair it yourself
and send me a diff). Note that it would need rather heavy patching
or it had to depend on gnu grep.
ok aja@
