Rubber <URL: http://rubber.sourceforge.net/> is "a program whose purpose is to handle all tasks related to the compilation of LaTeX documents." Reproducible: Always Steps to Reproduce: The code does have a bug related to passing commands to submodules. I'll attach a patch. The patch also changes the default lhs2TeX processing style from 'math' to 'poly'. 'poly' is a superset of 'math' and is where new features will end up in the future (in lhs2TeX), and seems to be the safest default.
Created attachment 27543 [details] ebuild for rubber
Created attachment 27544 [details, diff] patch to fix module command bug and to change default lhs2tex style Possibly the lhs2TeX change should be elided, or moved to a separate patch.
Created attachment 27789 [details, diff] new version of patch fixes bug in lhs2TeX module lhs2TeX module had a bug: when processing a .lhs file which was calculated to be a dependency, the module lost the name of the target .tex file. This patch replaces the earlier patch and fixes this problem. (Also sets --poly mode for lhs2TeX by default under all circumstances.)
Thanks for submitting the ebuild. It emerges and works fine. However, your ebuild doesn't have copyright header. We ask all submitters to assign ebuilds' copyright to Gentoo Technologies, Inc. due to intellectual property. If you are ok to do so, I'll include the ebuild to Portage tree. Also, SRC_URI should be SRC_URI="http://ebeffara.free.fr/pub/${P}.tar.gz" (It is much easy to update.) and virtual/tetex should be added to DEPEND. (in addition to these, use epatch rather than calling patch directly).
While I am all for making the --poly style the default in lhs2TeX calls, I think this is a fundamental change in the behaviour of the package. It should be suggested upstream, not fixed in Gentoo. Furthermore, if you fix it, it should most probably change "-math" into "--poly", not "-poly". The latter might work, but is deprecated. ks
0.99.8 was relaesed, so I created a new ebuild based on Sam's and added to CVS. Sam: I omitted the patch as suggested by kosmikus in comment #5, and would you send it to rubber upstream? Thanks in advance.
Created attachment 63634 [details] Ebuild for rubber-1.0 Hi, I adapted the ebuild from 0.99.8 to the new 1.0 version of rubber. Basically changed the name, and added dependancy to >=python-2.2 and virtual/tetex And I think 0.99.8 (or even 1.0) could be unmasked, too. Regards
I would like rubber-1.0 to be in the portage tree too. Let's try to wake up someone who can do that :) alexandre: maybe you shoud have opened a new bug for this.
You should open a new bug report... or reopen this one. Otherwise no developer will notice it! Btw. I could see on the rubber homepage that a zsh-completion function is shipping with rubber-1.0. Is it installed by default and/or has anybody tried it out?!?
Ah... It's not shipping with rubber but with the next version of zsh... sorry my bad!
I opened a new bug report 104405 for this. I took your ebuild Alexandre and attached to that bug report... but you got the credit :-)
rubber 0.99.8 (1.0 isn't yet in the tree, by lack of staff in text-markup) has been in ~x86 for a very long time now. No bugs were ever reported and I believe it received some testing (in my school, everyone uses it). x86 team : could it be stabilized, please ?