Hej David, If you think that RAM exaustion is going to cause problems, why don't you also add the recording of the memory useage in the doctesting (I wouldn't now how to exactly do this but I guess you could use the output of command line utilities such as top, vmstat/vm_stat or free). If this would make the output of "sage -t" to long for the average user/debugger you could always make a verbose and/or memory option to the doctesting which could be used by running "sage -tv" or "sage -tm" or or something alike.
I hope you find these ideas usefull. Kind Rigards, Maarten Derickx PS. and Dutch saying which I got from my father says: "Meet 't en je weet 't" wich translates to "Measure it and you know it". I think this saying also applies in this situation ;). On Jul 28, 12:13 am, David Kirkby <david.kir...@onetel.net> wrote: > On 27 July 2010 22:10, Jeroen Demeyer <jdeme...@cage.ugent.be> wrote: > > > > > > > On 2010-07-27 21:25, David Kirkby wrote: > >> Currently if a test fails in the doctest, one can never be sure if the > >> machine may have run out of swap space, disk space disk I/O problems > >> or all manner of other things that could cause test failures. But if > >> the time of the failure was recorded, then the system logs could be > >> used to see if anything was wrong at the time of the failure. > > > I have no problem with your proposal, but I don't see how it will solve > > your problem. I doubt that "2+2" is going to return "5" simply because > > of problems with the system. You probably get some exception saying > > that you're out of memory or that the test crashed or whatever. And > > manually checking the error looks easier than manually checking the time > > of the failure. Or am I missing something? > > > Jeroen. > > There have been reports of doc tests failing on sage.math when all RAM > has been exhausted. I've seen it on t2 too, where the machine has run > out of swap space. I think the issue of RAM exhaustion is getting > worst now its possible to build lots of packages in parallel. I > believe William implemented some limits on sage.math because of lack > of resources. > > There are very often doc tests which fail when one runs > > make ptestlong > > but pass when one runs the test on the command line. I have sometimes > suspected that it might be a lack of system resources. > > Does anyone else have any comments? > > Dave -- 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
