On Mon, Oct 8, 2018 at 7:53 AM Jeroen Demeyer <j.deme...@ugent.be> wrote: > > On 2018-10-07 21:32, Volker Braun wrote: > > I've seen this sometimes when running out of memory (docbuild gets stuck > > instead of erroring out) > > This is just general suckiness of multiprocessing.Pool > > I have a long-standing plan to extract the multiprocess code from the > doctesting framework and use that in other place like the docbuilder too.
Hahah, same. I had a prototype for this working more than a year ago but I got frustrated by some tiny bug somewhere and never went back to it. I should probably try again and start a bit simpler... It's not too hard to take the already existing code almost verbatim and just replace the body of the Worker.run method with something more generic. However, I was also trying to support the full concurrent.futures API (but extended to include the doctest framework's support for stdout capture from subprocesses), and that's where things got hairy. I should probably dig up that code and take another stab at it... -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. For more options, visit https://groups.google.com/d/optout.