You make it sound like there is just not enough doctesting coverage. The Sage doctests generally do not generate a lot of parents in one go. Maybe its just that the coverage of this use case needs to be improved? E.g. create a list of thousands of parents, delete random subset, garbage collect, repeat?
I admit that I haven't followed these patches as much as I would. Its clear that deleting parents can trigger lots of nasty stuff. We need to understand how to exercise that code. If we can agree to dedicating a point release to this issue then that just means that beta0 is going to be broken on some systems. I take it this is Nils' original objection: Not every beta has to work perfectly on every system. If you merge a hundred small patches then its reasonable to kick everything back out that triggers a doctest failure. But if you want to make progress on a big issue then you have to accept that a beta is going to be imperfect and meant to expose a ticket to a much wider audience. On Saturday, November 3, 2012 11:42:04 PM UTC, Jeroen Demeyer wrote: > > On 2012-11-04 00:18, Volker Braun wrote: > > I'd say talk to Jeroen to make collectable parents a priority for one > > release. For example, lets have 5.5 as a the release where we add the > > collectable parents. Push out a beta1 with these patches, then we'll > > have a month during Jeroen's holiday where we can check any other > > tickets. No other tickets get merged if they break the parents stuff. > An extra complication is that the breakage is often non-reproducible and > system-dependent. Together with the wierd interaction between seemingly > unrelated patches, even determining whether a patch breaks the parent > stuff is very non-trivial. > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. 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. Visit this group at http://groups.google.com/group/sage-devel?hl=en.
