William Stein wrote: > On Tue, Jul 21, 2009 at 9:51 AM, Dr. David > Kirkby<david.kir...@onetel.net> wrote: >> Sometimes when I create a patch, or look at work of others, I see >> potential issues that might come back to haunt us at a later date. For >> example. >> >> 1) Whilst updating a polybori patch, I noticed something done some time >> back, that I think might have an impact on Solaris. I don't have the >> time to check it fully. It would only happen if Sage is built in debug >> mode. It's hardly a critical issue, but I think it worth noting. >> >> 2) I'm applying a patch to pari, which involves commenting out a couple >> of lines of code that cause hassle on Solaris when building the >> 'modified sage library code'. >> >> The patch was suggested by William, it does work, but I think it is fair >> to say neither of us have really investigated the effect this might >> have. To do so would be huge amount of work, and probably result in the >> conclusion it can't hurt. > > I can't remember now exactly what I did, but I think that it was *not* > ever my intention that this actually get into Sage. It was just meant > to be a quick hack to get passed that point and finish building Sage. > One should figure out a correct fix by consulting with the pari > developers who are very very responsive.
The problem is, it is not the pari distribution that shows this problem. Pari compiles without any problems. The problem is in the 'modified sage library code', which includes this header file, causes the problem on Solaris. Realistically, it would be better if whoever wrote that library code discussed it with the pari developers, as they know why they included the file. Ignoring that specific issue, what is your thoughts in general about having some sort of trac ticket or other place where people list potential problems for the future? With the best will in the world, if a new issue of Sage was never released without some issues remaining, the rate of progress would drop to a crawl. >> 3) I was going to apply a patch to singular, which is a bit of a hack. >> It will work and will never cause any problems, but it is far from >> clean. A better solution would be to resolve the underlying problems, >> which are probably an autoconf bug or a bug in the singular build >> system. Someone else is looking into that one, but the fix is not going >> to be immediate. >> >> 4) An atlas patch will have a (I believe small) impact on Solaris sun4v >> systems, and if Sun can sort out the library, will no longer be needed. >> >> It seems to me there should be a central location where people can add >> to a list of potential problems that might just come back to bit us in >> the future. >> >> Perhaps one trac ticket 'Potential issues in X' and another 'Potential >> issues in Y' where anyone adds any potential issues they see, which >> might be a problem in package X or Y at some later date, but which are >> of low priority, or unlikely to occur. >> >> Dave >> > > > --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
