On Sunday, 26 April 2015 20:55:44 UTC+1, vdelecroix wrote: > > On 26/04/15 21:50, Dima Pasechnik wrote: > > On Sunday, 26 April 2015 20:32:57 UTC+1, Nathann Cohen wrote: > >>> I would advocate the approach that all the MILP tests are explicit > about > >> the > >>> solver to be > >>> used. Indeed, it's not only COIN, but all the other solvers like CPLEX > >>> behave this way. > >>> > >>> Nathann, would you mind this? > >> > >> HMmmmm... Why would you gain? If you make all doctests involving > >> solvers use GLPK, you only make sure that the other solvers will never > >> be tested anymore? > >> > > > > no no no, I mean to have different versions of doctests with outputs > from > > particular solvers, and these marked by # optional tags > > > > e.g. if your default solver is CPLEX then you'd doctest with > > --optional=sage,cplex > > and (ideally) should get no errors, etc. > > > > It introduces complications since nobody (?) uses > > sage -tp --all --optional=xyz > > I am using this (ok, call me nobody :-))
> Moreover, it is clear that by default you want the fastest solver > available. Imagine the number of functions using MILP in the Sage > library... all of them would use GLPK by default? there are not that many tests which are solver-specific; most of them are about printing the LP, which varies quite a bit from one backend to another. Perhaps I was not clear enough: I meant to say: make the solver-specific doctests work; we talk about a dozen doctests or so... Dima > > That being said, it would be a nice feature for patchbots to also tests > these optional doctests (on a reasonable base of optional packages). > > Vincent > -- 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.
