On 2010-12-22 15:39, Cedric wrote: > With regard to... > > 1.) I don't see a problem with that - the only problem might be the > rather rigid concept of doctest itsself. I agree that the doctests are rather rigid, but since they are very useful, I believe we should certainly not abandon doctests.
> But if the routines return > different values which are still mathematically correct and you can't > make doctest account for it then that's the way it shall be, no? I completely disagree, it would make doctesting useless. Checking whether a doctest is literally the same as a given output can be easily done by a computer, checking whether a different result is still mathematically correct is much harder. > 2.) How about submitting the patches upstream of the > specific package and wait for their commit? This is what we usually do. However, keep in mind a) upstream might be slow and we would like the patches now b) upstream might not agree with our patches c) some patches are very Sage-specific (for example, implementing the interface between a package and Sage) and it would not make sense to commit them upstream. > Summing my above counter arguments up I feel you tend to state the > exact same problems other software faces, too, and, too put it > provocatively, claim that SAGE can't be bothered to cope with these > common problems that lie in the nature of dependencies. I don't have a lot of experience with other software, so I can't really comment on this point. But also keep in mind that Sage has a *huge* number of dependencies. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
