On Jun 17, 2010, at 2:44 PM, Ryan Schmidt wrote: > > On Jun 16, 2010, at 19:31, Stephen Langer wrote: > >> Therefore it's a serious mistake for a packaging system to assume that it's >> ok to enable openmp in libraries. A quick solution would be to provide >> both openmp and no-openmp variants, which would make users choose between >> fast stand-alone ImageMagick programs and libraries that can be linked by >> threaded apps. > > We don't need two variants; we only need one variant, "openmp", which the > user can either enable or disable.
That's what I meant. I guess I was using the word "variant" in a nontechnical sense. > It just remains a question as to whether the variant should be enabled by > default or not. What I'm hearing is that we should disable it by default. That would break the least amount of code. > >> A better solution might be for the openmp and non-openmp versions of the >> libraries to have different names, so that both could be installed on the >> same system. > > Ugh. That sounds nasty. I agree. Can we get ImageMagick to allow openMP to be enabled or disabled at run time? That would also solve the problem. Such a switch doesn't exist at the moment, as far as I can tell. -- Steve -- -- stephen.lan...@nist.gov Tel: (301) 975-5423 -- -- http://math.nist.gov/mcsd/Staff/SLanger/ Fax: (301) 975-3553 -- -- NIST, 100 Bureau Drive, Stop 8910, Gaithersburg, Md 20899-8910 -- -- "I don't think this will work. That's why it's science." -- -- Naomi Langer (age 6), 17 Feb 2003 -- _______________________________________________ macports-users mailing list macports-users@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/macports-users
