I need to do some other things now, so will return to this later. But I 
believe I've found some of the reasons Sage is not building on 
t2.math.washington.edu whereas it does on my Blade 2000.

1) It is possible the gcc 4.3.1 binaries in 
/usr/local/sparc-solaris-toolchain were created in a newer version of 
Solaris than what is running on t2. According to some people on 
comp.unix.solaris, doing this could cause problems.

without knowing exactly what Michael used I can't be sure.


2) I tried building the latest gmp (4.3.1) using the Solaris toolchain. 
It built ok and passed the tests.

I installed that in

/usr/local/gmp-4.3.1-Solaris-10-u4-ABI=32

3) I downloaded the latest mpfr-2.4.1, patched it to the latest patch 
level (5) and built that, again using the compiler in 
/usr/local/sparc-solaris-toolchain. Whilst mpfr compiled ok, it failed 
about 20 tests of the 148 tests.

4) Now, somewhat suspicious that the tool chain is working ok on t2, I 
decided to have another go building gmp and mpfr, but this time using 
the Sun-supplied gcc (3.4.3) in /usr/sfw/bin.

This time the tests of both gmp and mpfr passed

I've put two directories in /usr/local

gmp-4.3.1-Solaris-10-u4-ABI=32-gcc-3.4.3
mpfr-2.4.1p5-Solaris-10-u4-gcc-3.4.3

which have the latest gmp and mpfr which pass tests. But as indicated on 
the directory names, these were built with gcc 3.4.3, not the Solaris 
toolchain, as that causes test failures in mpfr.

I'm trying to build gcc-4.4.0 now. I'll leave running while I do 
something else.

I know Micheal put a lot of work into getting the tool chain right, but 
i think it is failing on the t2, as I suspect he possibly built it on a 
machine with a newer version of Solaris.

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
-~----------~----~----~----~------~----~------~--~---

Reply via email to