Hi Nathann, Thank you very much. With your help I was able to compile the library from the "new" sources and it is working properly now, although just for me of course. I understand your disappointment because of not being able to solve the issue for everybody.
I am willing to write to the author of the C code but I do not know what exactly to ask him to do. The main problem is that the code works right in some architectures (at least in ours and in the one of the author I guess). So what is needed the most is active involvement from some other Sage user interested in having the modular_decomposition package to work in a different architecture so as detect what should be modified in the C code to make it work in that architecture too. Is there someone out there? I really appreciate your effort to make modular decomposition available in Sage because for some of us this is very useful. I think that keeping modular_decomposition as an optional package should be useful for testing purposes in different architectures until the portability problem with the source code is identified and solved. Thanks, Paulo On Saturday, May 2, 2015 at 12:33:11 PM UTC-3, Nathann Cohen wrote: > > Hello ! > > > I would like to know if there is any workaround for solving this issue > for > > version 6.6. Unfortunately, there is no dm.c and no random.c in the sage > 6.6 > > directory structure so as to replace them and so I do not know how to > > proceed. > > Sigh... Modular decomposition. A story of many disappointments. > > No, right now you will not find those two files in Sage's source code. > They are, however, contained in the optional package that is > downloaded when you run "sage -i modular_decomposition", and all that > is done on them is compile them into a shared library named > libmodulardecomposition (if I remember correctly. Not of my own > computer at the moment) located in SAGE_ROOT/local/lib/. If you > compile this shared library from the new sources, there should not be > any problem. If you do not know how, open the archive and look at > "spkg-install", it contains the lines that do that. > > I feel a bit guilty telling you how to make it work on your computer; > for we have a REAL problem with this package. What we need is to get > the original authors to solve it. > > 1) The old version of the sources (those that we ship) returns wrong > results. For instance on yours, as you were the one who reported it > first. > > 2) The new version of the sources apparently returns correct results > on your architecture (and on mine), but .... not on all of them. Look > at this ticket I opened when you first reported it: > http://trac.sagemath.org/ticket/13744 > > As you can see, the "new" version still triggers segfaults or wrong > results on some machines that we use to test Sage. Thus, my attempts > at upgrading Sage's version of the code were (rather legitimately) > refused. > > I am stuck with those files, and I do not like this situation (at > all). We ship something which returns wrong results, the problem is > fixed by updating to a new version which returns wrong results too or > even segfaults, and of course I get absolutely no answer from the > authors, whom I reminded regularly of the problem. > > To be honest, I feel like removing this from Sage. That was part of > the reason behind that more recent change, which made it an optional > package. > > Really, the best you could do to help us is send an email to the > authors. Tell them about your problem, and how cool it would be if it > worked, as this situation is bad advertisement for everybody. > Computing modular decompositions is cool and everything, but we can't > just keep buggy code in Sage, even if we raise a warning whenever it > is used :-/ > > Nathann > -- You received this message because you are subscribed to the Google Groups "sage-support" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-support+unsubscr...@googlegroups.com. To post to this group, send email to sage-support@googlegroups.com. Visit this group at http://groups.google.com/group/sage-support. For more options, visit https://groups.google.com/d/optout.
