On Thu, Jul 15, 2010 at 2:22 AM, John Cremona <john.crem...@gmail.com> wrote: >> >> +1 I really like this idea. All the advantages of mercurial queues >> without the late dependancies or requirement that the source be under >> revision control. (The patch queue itself would be under the .spkg >> control). The actual pushing of a series could probably be a simple >> command in spkg-install, if we wanted to make the spkg optional. >> > > Me too. I have just been scrutinising the spkg-install for the pari > spkg and find that there are 10 files in patches/ which are applied as > follows: > > * mostly they are just copied over the corresponding files in src/; > in some cases that deoends on the architecture > * one of them is plain copied if [ `uname` = "Linux" ] while if [ > `uname` = "SunOS" ] then a sed script is applied to edit the original > * two of them (header files) are only replaced after pari is built > since the originals are needed to build it but changed files are > needed for the Sage library. > > That is rather complicated (and I'm sure there are worse). And it is > a real pain to update the src -- one of those header files has changed > a lot even in the last 2 weeks since William and Robert made > pari-2.4.3.svn.p3 and so I need to do some manual editing to get their > changes into the p4 I am making; changes which I cannot test since > they are only needed on Darwin or SunOS. > > In particular, any new system we put in place has to have conditional > application of patches depending on operating system.
Are there any packages where #ifders and/or environment-sensitive makefiles would not do the job? - 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
