Steve, unresolved symbols in library is a "feature" and yes it can make linking dodgy, as here when one linker needs some stuff that is not there to give you a viable result.
Sorry to Have implicated you Jean-Pierre I just saw that your cygwin patch touched cblas in gsl so there is a slim chance that it altered the linking of the library. But really it is probably an upstream feature so that you are not locked in using libgslcblas but then you have to know what you are doing when linking. I am not sure which option to pass to gold to make the problem go away got OP to try --no-undefined on complex_double.so but it doesn't seem to have worked. Unless I got the syntax wrong of course. Francois ________________________________________ From: sage-devel@googlegroups.com [sage-devel@googlegroups.com] on behalf of Steven Trogdon [steven.trog...@gmail.com] Sent: Saturday, 8 February 2014 8:10 To: sage-devel@googlegroups.com Subject: Re: [sage-devel] Error installing package conway_polynomials-0.5.p0 on Ubuntu 12.04 - Sage 6.1.1 On Friday, February 7, 2014 12:31:59 PM UTC-6, François wrote: While gold is part of the problem it is not completely its fault in my opinion. gold I think default to "as-needed" sage/rings/model. complex_double.pyx doesn't directly call cblas so gold doesn't see it as needed and drop it. The problem here is that libgsl itself is not linked properly for it to work - regardless of gold. If you check the readelf -d output for libgsl there is no mention of libcblas/libgslcblas in the nedded section. So the libgsl doesn't even know where to find its undefined symbols. I guess gold may be made to work here if we can convince it to include libcblas in the needed section of complex_double.so. But I think we should fix the somewhat fishy stuff in libgsl, sure it is legal and all in that you can link with the libclas of your choice and gold may even do the right thing for an executable where all symbols have to be resolved but a .so? Well undefined symbols is legal for a library..... Yes, I do find it somewhat odd that even with a properly built and linked Sage there are undefined symols in libgsl.so. There must be some level of linking where even this will fail. This email may be confidential and subject to legal privilege, it may not reflect the views of the University of Canterbury, and it is not guaranteed to be virus free. If you are not an intended recipient, please notify the sender immediately and erase all copies of the message and any attachments. Please refer to http://www.canterbury.ac.nz/emaildisclaimer for more information. -- 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 http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/groups/opt_out.
