On Thu, Oct 29, 2015 at 10:42 AM, Nathann Cohen <nathann.co...@gmail.com> wrote: >> That's what >> >> # long time >> >> is supposed to be for. > > I don't think so. You wouldn't want to see a doctest take one hour to > run, even if it is flagged with 'long time'. I don't know about you, > but I run 'long' doctests several time per day on some files. > >>> The same happens for some doctests that rely on optional LP solvers: >>> they can take <1s with CPLEX/Gurobi, but 'standard sage' (with GLPK) >>> cannot be expected to ever solve them. >> >> So if a user tries to use them they will be effectively broken? > > Broken? Why? They do return the expected output, they only do it very > very slowly (and perhaps not at all) when you don't have "one of > several solvers" installed. When you do, it can be instantaneous. > Thus, it is probably better to not test them every time somebody types > 'make ptestlong'.
There is no place for tests that cannot be automatically run somehow in our test suite. Absolutely any instance of this whatsoever is a shortcoming in our test suite that we should *plan* to find a way to address. (I'm of course not saying we should delete/change everything right now.) If you have tests that take too long, then we need another tag "# really long". If there are tests that create git branches or have other "undesirable side effects", then get better at writing tests that properly run in a sandbox that addresses such side effects. William -- William (http://wstein.org) -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
