On Sun, Sep 21, 2008 at 12:04 PM, Georg S. Weber <[EMAIL PROTECTED]> wrote: > > On Intel MacBook OS X 10.4 / Xcode 2.5 (yes, I upgraded), nothing > really new for 3.1.3.alpha0: > > builds fine > > make test choked on two of three runs on /modular/abvar/homspace.py, > and passed on the third > > make testlong choked on sr.py and ell_finite_field.py (the known ones > from the 3.1.2 cycle) > > > The "homspace.py" doctest takes quite a lot of time, more than 240s == > 4 minutes in the best case, test "long" or not "long" making no > difference (on 2 GHz ...). > > Is that essentially OK? > > Somehow, I still haven't found out under which circumstances exactly, > OS X uses --- of the two CPU cores of the Core2 Dual --- only one core > for the (build and) doctests. If so, "homspace.py" takes well over > 400s, passes (always, without problem) the "testlong" testsuite, while > the "short" testsuite throws a "TIMED OUT" failure. > > I didn't look at the code yet, but I guess that for the usual "short" > testsuite, there is some timeout mechanism enabled, which gets > disabled for the "long" testsuite. The latter runs only for those > doctests longer, for which extra "long" doctests are defined, of > course. > > > Essence: > Should I open a trac ticket for the "modular/abvar/homspace.py" > doctest to be changed to be a "long" doctest (i.e. the biggest part of > it)? > > > Or will that probably go away after some more merges of the new > coercion code? (Again, I didn't look at the code whether this could > make sense.) >
Coercion won't fix this. I think some doctests should be changed to #long in that file, and -- better -- Craig Citro and I will get off are butts and come up with better/faster algorithms to make it so all those doctests run much more quickly. That it takes as long as it does now is I bet just the result of not good enough algorithms and/or implementations. William --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
