On Thu, Jan 19, 2017 at 8:47 AM, Frédéric Chapoton <fchapot...@gmail.com> wrote: > Hello, > > let me try to summarize : we are still quite far from being python3 > compatible. > > There are two main milestones we must reach: > > 1) being able to build sage with python3 ; > (with substeps: 1A) being able to cythonize all pyx files / 1B) being able > to compile all pyx files) 1C) being able to build all py files. > > 2) having all doctests pass > > Currently, progress is very slowly being made on 1) and some changes are > made in parallel (in a blind way) for 2) > > Concerning 1), I would say cmp and unicode are the main remaining problems. > Integer division has almost been tackled, but not quite. > > The problem with cmp is quite hard, and it seems that every instance of > cmp() must be handled separately. I have gained some experience on the > subject, but given the lack of reviewers, this implies a very slow pace. > This cmp problem should be the last remaining thing to handle before > reaching 1A) and maybe even 1B) > > For 1C), the cmp problem has not been considered yet, but it is huge. > > Unicode problem has almost not been looked at either. I just had a quick > look at #14153. > > Concerning having doctests pass 2), some work has already been done, but > given that we are not able to build, this is done without any check, which > is bad. Here, we are trying (in the dark) to prepare things for when build > will happen. > > Some general comments : I have been told to make progress by small steps, so > using many tickets. This is maybe better for stability, but implies that > every single move takes a long time to enter sage: review and closing may > take months. Very often, I must stop working on python3, because there are > already many positively reviewed tickets, and only much later they willl be > closed. I need them to be cleaned up to have a clear view of the situation. > > Another problem is regression. I have made patchbot plugins to try and > prevent incompatible code to enter sage again. This is not sufficient. We > need to have the doctest framework itself check that there is no bad print > and a few other solved issues. I have required help to do that, but nobody > stepped in. > > I would like to thank here the people that have been tirelessly refering so > many of the python3 tickets, in particular Travis, Jeroen and Jori. If some > other people are interested, reviewing tickets would be helpful. Or even > working on the remaining issues. I did not want to do all this just me > alone...
I am hoping to help more with this especially once the Cygwin situation is wrapped up a bit more neatly and I have the Cygwin built-bot up and running. I can especially help with string-related issues as I have a lot of experience with Python 3 porting, and especially as it relates to that. I can't be of much help with the cmp stuff since it seems to rely on knowledge of how various objects in Sage are expected to work that I don't have a firm grasp on. As you say, it needs to be handled on a case-by-case basis and it would probably be a big help to you (and Sage in general) to have a bugbash on that. -- 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 https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
