On Tuesday, March 19, 2013 7:51:42 AM UTC-4, Stephen Montgomery-Smith wrote: > > On 03/19/2013 06:07 AM, leif wrote: > > Jeroen Demeyer wrote: > >> On 2013-03-19 04:30, John H Palmieri wrote: > >>> Is there any good way of testing whether multithreading is broken > >> That would of course be ideal, but we should first find out exactly > >> why/how it's broken on FreeBSD. > > > > Yep. As said before, apparently the Sage library /did/ build (which > > *always* uses multiprocessing, even if SAGE_NUM_THREADS=1, i.e., even if > > it uses just one thread, AFAIK). > > > > I also wouldn't special-case on FreeBSD in the docbuilder, but use > > SAGE_DOCBUILD_NUM_THREADS (say) there instead, and set it to 1 if > > necessary elsewhere, otherwise let it default to SAGE_NUM_THREADS or > > whatever. (The user should be able to set it as well.) > > I should add that the doc building for FreeBSD breaks, even if the > number of threads is 1. It is the use of Pool that seems to be the > problem. > > I don't think this problem can be tested by seeing if threading works, > because I don't know a reliable test. I write small test programs, and > they seem to work just fine. > > For all I know, it works if the build is on a 386, and not on a 64 bit > machine. > > The breakage is very mysterious, and is probably some horribly difficult > to test race condition. >
My understanding is that you did eventually fix this, correct? I assume so from http://www.freebsd.org/cgi/cvsweb.cgi/ports/math/sage/files/spkg-patch-sage_-_doc_common_builder.py?rev=1.1;content-type=text%2Fplain Did anyone ever open a ticket for this? I can but don't want to duplicate. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
