On Sat, Apr 18, 2009 at 6:34 PM, mabshoff <mabsh...@googlemail.com> wrote: > > > > On Apr 18, 6:06 pm, Alex Ghitza <aghi...@gmail.com> wrote: >> Hi sage-devel, > > Hi Alex, > >> I'm slowly and painfully making my way through >> schemes/generic/morphism.py and have run into some trouble that >> persisted even after looking at rings/morphism.pyx which was recently >> doctested by William. >> >> So here are some questions: >> >> 1. sage -coverage asks for a doctest of the form s == loads(dumps(s)). >> As far as I can tell however, it is happy if there is one such test >> in a file, even if the file happens to define 20 different classes. I >> would assume that we want a doctest of this form for each class >> definition, is that right? If so, can/should sage -coverage be >> changed to indicate this? > > Yes, I think that is a bug and not having pickles of all classes will > cause trouble down the road. Please open a ticket.
It definitely would be nice if this functionality were added. > >> 2. at the moment, s == loads(dumps(s)) fails miserably for any of the >> classes in schemes/generic/morphism.py. The problem is in the >> comparison, and the classes do not define __cmp__() methods. I also >> noticed that the class they inherit from, Map, also does not implement >> its own __cmp__(). So the question is: am I right in assuming that I >> have to implement __cmp__() in my classes? Would it be a good idea to >> also implement a generic __cmp__() for Map, or is there a good reason >> why this was left out? You need to define __cmp__. The default __cmp__ is basically just "is". Part of the value of s==loads(dumps(s)) tests is that they also test __cmp__. >> 3. as I mentioned, I looked in rings/morphism.pyx for inspiration; >> this helped, but not enough so I tried >> sage: f = ZZ.hom(QQ) >> sage: g = loads(dumps(f)) >> sage: f == g >> False >> >> Shouldn't this return True? Yes >> Shouldn't a test like this be somewhere >> in rings/morphism.pyx? Yes. > > I am not sure about the last two questions, so I will wait for someone > more knowledgeable to answer this. > >> It would be great to know more about the issues involved here. I'm >> really not familiar with this part of the Sage infrastructure and it >> shows :) > William --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
