compiler.
Add a patch to let pdftoepdf from luatex to still build (found by
sthen@).
I hope this will also help architectures without clang in base
(where this port is currently broken).
Should have no ill effects on archs with clang in base, and I did
some light testing before the switch of amd64 and i386 to clang
(when both COMPILER_LIBCXX and LIBCXX were used) without hitting
any problem.
Tentative ok from edd@, a long time ago.
In some cases there still happen -bindir-related failure, but I failed
to reproduce, and noone else sees them for now. Even more, those two
patches do not cover all -bindir cases in sources, so if/when libtool
will fail, you'd need two more patches anyway.
okay aja@
texlive_texmf-buildset, so we can build all the documentation
(vignettes, html, etc.) for math/R.
Much help and ok edd@ (aka "the TeXmeister")
OK and help from sthen@ too
This makes it easier for other developers to move things between the texmf
subsets. For example when adding things to the buildset.
OK landry@, Thanks!
Although latexmk is included in texlive, we omit it in favour of our
separate port. texlive_base should not install a symlink.
There was also confusion between latex-mk and latexmk. We have both in
ports, but the latter is the one related to texlive. Let's mark is as a soft
dependency.
Upgrade path is currently leaving a partial package behind due to the bogus
symlink. A quirk will probably be required, which we can commit later.
OK ajacoutot@
This started as a fix (arm, ppc) for the internal icu in texlive. It turns
out a custom icu is not needed anymore, so we can use the icu in the ports
tree. This makes handling of header files much simpler since there is no
longer the possibility of picking up the wrong icu headers. Should also fix
PPC and ARM builds of texlive.
Also noticed a few other libraries being built into texlive. Fixed.
Finally, readme was not being installed. Fixed.
Reported by landry@
OK dcoppa@, landry@, and with input and testing from kili@.
1) If the user updates only the buildset, ensure hashing occurs.
2) When I moved stuff from PLIST-main to PLIST-buildset, I should have
updated a conflict marker and added a tighter RUN_DEPEND. This ensures
that the two packages affected are updated together.
OK sthen@
Now that the last of the ports which was installing in the deprecated
TEXMFMAIN have been fixed, the TEXMFMAIN texmf tree can dissappear. We do
this by falling in line with the upstream configurations.
OK naddy@
many were in pre-formatted form in other packages so they didn't actually
conflict previously even though the PLIST entry was identical - recently
a number of these are now installed as unformatted pages which has caused
a real conflict. No substantial diff between texmf's versions and the
other preformatted pages (minor whitespace changes due to groff/mandoc
differences, that's all).
for consistency with the gem/extconf ports.
Update non-gem/extconf ports that relied on the previous
default of ruby 1.8 to explicitly specify it.
OK ajacoutot@
Many people have mailed me successful test reports and matthieu tells me he
has been running this for some time now.
Thanks to those that tested.
OK matthieu@
