hmm, on Tue, Feb 04, 2014 at 02:08:24AM +0100, Juan Francisco Cantero Hurtado said that > On Mon, Feb 03, 2014 at 09:07:51AM +0100, frantisek holop wrote: > > hmm, on Sun, Feb 02, 2014 at 11:05:51PM +0100, Juan Francisco Cantero > > Hurtado said that > > > The latex support in sphinx is important, so remove texmf from test deps > > > is not a good idea. frantisek, I know texmf is really big but you only > > > need download the distfile one time per year. > > > > this is not about latex support in sphinx. > > some sphinx _tests_ use texmf, it's for > > sphix developers to make sure they dont > > break something when they make changes > > to the latex support. > > > > Sorry, I wasn't clear in my email. We use the regression tests in the > packages to test if everything works correctly. These are also used when > someone update some problematic dependency. If some test requires latex, > the dependency is mandatory. > > If you read the CVS log, I was the user who requested the change > to add latex to the test deps :)
ah, sorry i misunderstood. well, the usage of tests is somewhat liberal in my experience.. i have seen plenty of ports setting NO_TEST=Yes and not bothering with the TEST_DEPENDS and the tests and i dont blame them... ultimately, they are for the developers, and many of them dont include them in their release archives and/or dont include some crucial file(s) for the tests to be able to run (like tox.ini). some even consider the generation of docs (using sphinx) as a developer only task doable only from a git checkout :( but of course, i like to have tests (and pass them). having said that: - i am not going to install texmf - i dont mind keeping the dependency there, but someone else will have to raise the issue if latex tests are failing. -f -- 1st law of thermodynamics: go to class!
