Le vendredi 18 mai, Jeroen Demeyer a écrit: > On 2012-05-18 10:14, Julien Puydt wrote: > > The second point has bothered me for quite some time already, so > > with the first in the air, it might be time to ask : wouldn't it be > > worth to have a proper boost spkg, which would contain the needed > > frameworks, coming from upstream? > In principle I have nothing against including boost, but you should > know that it takes close to 50MB, which would make it the *largest* > package in Sage, apart from the Sage library itself. >
Well, the upstream boost_1_49_0.tar.bz2 is indeed about 50M, but : - it contains all the documentation ; - it contains the tools to build a few things (notably the documentation) ; - it contains all of boost, while the only thing using boost in sage, polybori, probably only uses only parts of it (which ones? Probably the unit test and the python). In fact, if you take the upstream tarball, and keep only the boost/ directory, you get : $ ls -hl boost_1_49_0_only_boost_dir.tar.bz2 -rw-r--r-- 1 jpuydt jpuydt 5,4M mai 18 12:08 boost_1_49_0_only_boost_dir.tar.bz2 As boost is mostly a pure-header library, this is probably enough (the current boost_cropped spkg-install only uses 'cp' -- no compilation, so it's not unreasonable). Would an SPKG.txt stating : "Packaged by taking upstream's full boost_1_49_0.tar.bz2 and keeping only the boost/ directory." and an spkg-install just doing cp be ok? Snark on #sagemath PS: Perhaps I can compile enough of sage-5.0 to see if that would fly with polybori... but I'm not sure. I find it annoying to propose things I can't test, but I don't think I can do better at the moment :-( -- 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
