Opened 2 years ago
Closed 2 years ago
Last modified 2 years ago
Gave it a first try on an 8.4 system, using the exact same instructions as we currently use for 2018, and make -j1 (because I had poppler problems failing to build the prerelease source).
Chugged on quite a long way, but failed in dvisvgm, undefined references to gsapi_exit, gsapi_new_instance and others. This seems to be similar to https://tug.org/pipermail/tlbuild/2018q3/004291.html (a long thread, explaining why TL will not be fixed for system libs in dvisvgm). But the patch proposed there for the dvisvgm Makefile.am will not apply, the current texk/dvisvgm/dvisvgm-src has nothing like the code it changed.
Looks as if -lgs (and perhaps -lpotrace) need to be added, but I'm out of my depth. Looking at distros, I know a SuSe dev was keen to be ready, I thought that was for OpenSuSe but that seems to still be on 2018. Arch have 2018, flagged as out of date. Gentoo don't seem to have anything after 2017.
I've come back to this: adding --disable-dvisvgm to our current instructions allows it to build, and no failures in 'make check'. Looking at dvisvgm, it converts DVI, EPS, PDF to SVG.
https://dvisvgm.de/ Doing without that (and perhaps eventually adding it as a separate package) is not ideal, but if it lets the bulk of it work that would be a step on the way.
Will give it a test.
Looks ok for the basic stuff. I was going to put the binary on another machine for basic tests, but there was a mail on tlbuild last night: https://tug.org/pipermail/tlbuild/2019q2/004526.html
AFAICS, updated binaries have not yet percolated out (if they have even been prepared).
Meanwhile, I've checked out svn://tug.org/texlive/branches/branch2019/Build/source/
and prepared a diff from -r 50882:51236
Will test that, although I don't think any of the changes will affect my test cases.
I think I've fixed this at r21647, but I've again been up all night so there are probably typos or errors in it.
I was reviewing r21647, and discovered that the upstream patch in the command explanations is referring to -31 instead of -1. Also, did you mean to leave the MD5SUM for the texmf file blank?
Powered by Trac 1.5.3.dev0
By Edgewall Software
© 1998-2021 Gerard Beekmans.