one way or the other, underlinking does not work on Cygwin. On Friday, June 2, 2017 at 9:54:22 AM UTC+1, Ralf Stephan wrote: > > No. See > https://stackoverflow.com/questions/44322187/binary-using-both-python-c-api-version-2-and-3 > > On Friday, June 2, 2017 at 8:35:57 AM UTC+2, François wrote: >> >> >> > On 2/06/2017, at 17:57, Ralf Stephan <gtr...@gmail.com> wrote: >> > >> > On Thursday, June 1, 2017 at 12:29:23 PM UTC+2, François wrote: >> > .. Installing pynac for both python at the same time >> > means rethinking its packaging. >> > >> > As a possible alternative, would avoiding those calls that differ >> > between version 2 and 3 of the C API help? It might not be so >> > difficult to do that. >> > >> >> Yes, it would, and in that case we would want to underlink >> libpynac. By that I mean not link libpynac to libpython when >> you build and install it so that when you link an application >> you can link with any python providing the right symbols. >> >> That’s what upstream gsl does with regards to cblas. >> >> The approach has some downsides. It doesn’t play well with >> the gold linker when you specify `—as-needed` at least. Which is >> why I fixed gsl in sage to be linked to the current cblas >> instead of leaving it as shipped upstream. >> However there could be ways to deal with that. >> >> The next issue is that I don’t know how well it works on >> OS X and cygwin. I am presuming it would work on OS X but >> I cannot make such statement for cygwin. >> >> A bonus point for that approach is that it would lead to >> the removal of all the pragma checking for python3 in the code. >> >> François >> >> >>
-- 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 https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
