Sourav recently experienced this same problem on sage.math -- I figured it was because the sage build failed, but now I'm thinking that a bug may have been introduced when dsage was moved into a spkg.
On Fri, Jun 5, 2009 at 11:54 AM, William Stein<wst...@gmail.com> wrote: > > 2009/6/5 Dr David Kirkby <drkir...@gmail.com>: >> >> >> >> On Jun 5, 8:48 am, Dan Drake <dr...@kaist.edu> wrote: >>> On Fri, Jun 5 2009 12:37 am, William Stein wrote: >>> > Have you got an account on t2.math.washington.edu yet? If so, it >>> > would be *extremely* useful if you could build Sage there (I haven't >>> > yet!), since then I could directly debug this notebook problem. >>> >>> Note that right now, building Sage fails on that machine with an >>> internal compiler error -- the Fortran compiler fails. There's a partial >>> build in /scratch/drake if you want to see an install log. >>> >>> Dan >>> >>> -- >>> --- Dan Drake <dr...@kaist.edu> >>> ----- KAIST Department of Mathematical Sciences >>> ------- http://mathsci.kaist.ac.kr/~drake >>> >>> signature.asc >>> < 1KViewDownload >> >> Just to second that Dan's comment, when I tried to build Sage >> sage-4.0.1.alpha0 on the T2, it failed for me too, with the Fortran >> compiler's internal error. >> >> The same compiler works fine on my Blade 2000, which is a dual >> processor box. >> >> I have in the past noticed some code that works fine on single >> processor machines fails when there are more processors - Mathematica >> has been a wonderful example of that, but I've noticed it on other >> code too. >> >> My Blade 2000 has two 1200 MHz processors, but that is nowhere near 8 >> cores. >> >> It might be worth a quick test of disabling the other cores, then >> trying the build that bit that uses fortran again. If William or >> someone else with root access could do that with the 'psradm' >> >> http://www.uwm.edu/cgi-bin/IMT/wwwman?topic=psradm(8)&msection= >> >> Does anyone know what sort of machine the tools were actually built >> on? Was it the T2 ? > > I think it was built on a Blade 2000 by Michael. I just tarred up his > directory and extracted it on T2. > >> >> I can think of a few possible things which might just get this to >> work. >> >> 1) Download the latest gcc and tried to build that on the T2, in the >> hope the error might not exist in the latest one. (best option). I'll >> do that later. > > That would be optimal from my point of view, since we need to port > sage to work with gcc-4.4.0 anyways. > > William > >> >> 2) Try to build sage with only one core enabled. That should only take >> 5 minutes to test, but needs root access. After putting the other >> cores offline, su - kirkby, change to the directory /home/kirkby/ >> Solairs/sage-4.0.1rc0 and type make again. See if it builds with one >> core. Remember to enable the other later afterwards. >> >> 3) I try to build the same version of the compiler, using the exact >> same options to the configure script that Micheal used, but do it on >> my Blade 2000. Then upload that to the T2 and try that. >> >> 4) I copy the relavant bit of the fortan binary of Sage built on my >> Blade over to the T2. That might get the Sage build to complete on the >> T2, as it build fine on my Blade 2000. >> >> I've never found the gcc developers too helpful on Solaris specific >> bugs. I can imagine them being even less so on Fortran/Solairs bugs. >> >> Dave >> >> >> >> >> >> >> > >> > > > > -- > William Stein > Associate Professor of Mathematics > University of Washington > http://wstein.org > > > > --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
