On Tue, Aug 24, 2010 at 7:47 AM, David Kirkby <david.kir...@onetel.net> wrote: > On 24 August 2010 15:39, David Kirkby <david.kir...@onetel.net> wrote: > >> Robert, >> >> Are you suggesting (1) or (2) below? >> >> 1) Run if tests take < 30 seconds AND <25% of the build time. >> >> 2) If tests take <30 seconds OR < 25% of the build time. >> >> I can see a problem with the latter. > > I mean I can see a problem with both! > > I think the 30 s on sage.math is a reasonable criteria alone. One > might argue for one minute, but I don't think the criteria needs to be > any more complex than just a simple time. > > You could make it more complex if you wanted, but then you would need > to consider exceptions like ATLAS, for the reasons I stated.
I wasn't proposing a rule, just a rule of thumb. And I did mean OR. If a package took 5 seconds to build, then 30 seconds to test, that would not be an issue for me, as relative to the total build time it's still negligible. OTOH, I wouldn't want to rule out a more expensive test for a more significant chunk spkg just because it broke the 30 second limit. Essentially what I'm saying is that, within reason, I'd be willing to give slightly more time to larger packages than smaller ones. ATLAS is a clear exception to the 25% idea, and I'm sure not the only one. - Robert -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org