As some of you know, I downloaded the Solaris toochain and managed to build sage on my Blade 2000. But Sage will not build on t2.math.washington.edu, due to a problem with gfortran.
The Blade 2500 on which the toolchain ws build is quite a similar machine to my Blade 2000, but t2 is quite different. However, t2 *should* run any binary created on the Blade 2500. I copied the toolchain to an older Ultra 60 of mine (2 x 450 MHz) and it fails to work on my U60 either. The error messages are different - even gcc fails on my Ultra 60 So t2.math.washington.edu does not appear to be unique in its inability to use the toolchain - nor can my U60. (My Ultra 60 happens to have Solaris 10 update 4, which is the same as t2.math.washington.edu, which might be relavant.). You may recall my recent thread '3.4.1 - 'mabshoff' hard-coded in 'linbox' http://groups.google.com/group/sage-devel/browse_thread/thread/e7380849a20c8256/fafbd98e7e7796ac?lnk=gst&q=mabshoff+solaris+david#fafbd98e7e7796ac where I reported an issue with errors like: can't open /home/mabshoff/sparc-solaris-toolchain//lib/libstdc++.la Micheal said to change the paths in sparc-solaris-toolchain//lib/libstdc++.la and other files ending in .la in the directory sparc-solaris-toolchain//lib/ I just noticed on t2.math.washington.edu that the paths are hard-coded to /home/mabshoff/ too. I fixed those (renaming the original files with the extension .orig), but it has not made any improvement - gfortran still fails. I'm hoping to build, test and install gcc-4.4.0 properly. While I have succeeded in building gcc-4.4.0 with no errors, the installation of gcc-4.0.0 was not perfect- I had to get make to skip some errors during the install process. I was unable to test gcc-4.4.0 since autogen does not exist on t2.math.washington.edu. Autogen happens to be one of those annoying programs which has several dependancies, all of which have their own dependencies - i.e. installing it is a pain in the butt. I did manage to use gcc-4.4.0 to compile part of Sage, but it failed when trying to compile pari. However, I don't really trust the gcc-4.4.0 I built, as the installation was not flawless and I can't test it. I'll look at this more in the next couple of days, but I will be tied up quite a bit on Monday. Clearly this is a non-trivial process. Every time i take a stop forward, it appears I need to go back two steps! 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 -~----------~----~----~----~------~----~------~--~---
