On Sun, Apr 3, 2016 at 10:32 PM, Dima Pasechnik <dimp...@gmail.com> wrote: > > > On Sunday, April 3, 2016 at 8:30:41 PM UTC+1, Erik Bray wrote: >> >> On Sun, Apr 3, 2016 at 4:45 PM, Dima Pasechnik <dim...@gmail.com> wrote: >> > IMHO the pre-reqs needed to build Sage on Linux ought to include >> > gfortran >> > and g++ (why not? listing them won't harm, even if the gcc version is >> > blacklisted) >> > >> > This is becoming an issue - there was a post recently about a failure to >> > build Sage on a newish Linux system, with error in building Sage's gcc >> > using >> > a new gcc 5. >> >> On the contrary, I believe requiring someone to build an entirely >> separate compiler toolchain is more burdensome and both user- and >> developer-hostile compared to fixing the issue with that compiler >> (where possible; I recognize it is not always). > > > Well, that's what I meant - our documentation does not spell out details on > when > one can use the system's default toolchain, and what are the requirements > for this. > > What happens is that sometimes people just did not install system's gfortran > and g++, > whereas gcc itself is OK, and this results in Sage building its own > toolchain, while > this could be avoided by installing gfortran and g++.
Ah okay, yes, definitely. For a while when I was building the Docker images it was compiling gcc for exactly this reason--I did not have gfortran. In fairness this requirement *is* documented, but it's not very specific. It's also hard to find in the build log--if possible at all--any note like "gfortran not found, building gcc". I'd almost rather the build exit at this point unless I've manually provided some flag (which of course would be pointed out in an exit message). But either way. A table somewhere in the documentation would help track this...? >> > On Sunday, April 3, 2016 at 9:42:58 AM UTC+1, Volker Braun wrote: >> >> >> >> Pretty clear that its a dual abi issue, libfplll discovers your system >> >> libqd and decides to link against it. But your system is new enough to >> >> use >> >> the new cxx11 abi, whereas the Sage-built gcc is not. Hence linking >> >> fails >> >> with undefined std::__cxx11 symbols, as expected. >> >> >> >> This is probably going to become a big issue for us soon when more >> >> distros >> >> switch to the new cxx11 abi. I've opened >> >> >> >> http://trac.sagemath.org/ticket/20350 >> >> >> >> A workaround for now would be to uninstall your libqd-devel package >> >> while >> >> compiling libfplll, or prevent Sage from building its own gcc (you >> >> probably >> >> don't have gfortran installed) >> >> >> >> >> >> >> > -- >> > 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+...@googlegroups.com. >> > To post to this group, send email to sage-...@googlegroups.com. >> > Visit this group at https://groups.google.com/group/sage-devel. >> > For more options, visit https://groups.google.com/d/optout. > > -- > 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. -- 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.
