>>>>> "td" == tim daly  <[email protected]> writes:

td> What operating system and version of OS are you using?
td> I'll build a machine and try to recreate it.

Its an amd64 gentoo box.  I built from gentoo's ebuild, updated to point
at the 201107 version.

Having saved the build tree, I linked that into its build-time location
and ran:

:; env AXIOM=/var/tmp/portage/sci-mathematics/axiom-201107/work/axiom/mnt/linux 
PATH=/opt/axiom/bin:$PATH /opt/axiom/bin/AXIOMsys

That worked fine.

Leaving out the link and pointing AXIOM directly to the location to
which I copied the build tree also worked:

:; env AXIOM=/home/cloos/axiom-201107/work/axiom/mnt/linux 
PATH=/opt/axiom/bin:$PATH /opt/axiom/bin/AXIOMsys

So I compared the .o files in .../work/axiom/mnt/linux/algebra vs the
ones installed into /opt/axiom/algebra.  They differ.

Something called by make install changes them.

I'll try again with gentoo's nostrip FEATURE selected to see whether
that makes the difference.

It'll take a few hours.  The tests take forever to complete.

Which reminds me: the failed tests summary is:

int/input/ApplicationProgramInterface.regress:regression result FAILED 2 of 5 
stanzas file ApplicationProgramInterface
int/input/DoubleFloat.regress:regression result FAILED 2 of 13 stanzas file 
DoubleFloat
int/input/GuessOptionFunctions0.regress:regression result FAILED 1 of 1 stanzas 
file GuessOptionFunctions0
int/input/PlaneAlgebraicCurvePlot.regress:regression result FAILED 2 of 5 
stanzas file PlaneAlgebraicCurvePlot
int/input/cachedf.regress:regression result FAILED 1 of 8 stanzas file cachedf
int/input/dfloat.regress:regression result FAILED 3 of 10 stanzas file dfloat
int/input/dftrig.regress:regression result FAILED 233 of 672 stanzas file dftrig
int/input/dop.regress:regression result FAILED 8 of 127 stanzas file dop
int/input/e1.regress:regression result FAILED 4 of 7 stanzas file e1
int/input/ei.regress:regression result FAILED 16 of 20 stanzas file ei
int/input/elemnum.regress:regression result FAILED 16 of 50 stanzas file elemnum
int/input/en.regress:regression result FAILED 5 of 7 stanzas file en
int/input/gamma.regress:regression result FAILED 4 of 12 stanzas file gamma
int/input/grpthry.regress:regression result FAILED 4 of 68 stanzas file grpthry
int/input/guess.regress:regression result FAILED 62 of 102 stanzas file guess
int/input/ico.regress:regression result FAILED 11 of 65 stanzas file ico
int/input/newtonlisp.regress:regression result FAILED 2 of 14 stanzas file 
newtonlisp
int/input/numericgamma.regress:regression result FAILED 7 of 36 stanzas file 
numericgamma
int/input/r20bugs.regress:regression result FAILED 1 of 27 stanzas file r20bugs
int/input/richalgebraic000-099.regress:regression result FAILED 2 of 518 
stanzas file richalgebraic000-099
int/input/richalgebraic100-199.regress:regression result FAILED 141 of 512 
stanzas file richalgebraic100-199
int/input/richalgebraic400-461.regress:regression result FAILED 4 of 324 
stanzas file richalgebraic400-461
int/input/richhyper1000-1098.regress:regression result FAILED 4 of 531 stanzas 
file richhyper1000-1098
int/input/richlog300-391.regress:regression result FAILED 164 of 460 stanzas 
file richlog300-391
int/input/richtrig000-099.regress:regression result FAILED 31 of 520 stanzas 
file richtrig000-099
int/input/richtrig800-899.regress:regression result FAILED 165 of 526 stanzas 
file richtrig800-899
int/input/tuplebug.regress:regression result FAILED 1 of 1 stanzas file tuplebug

Most of the test failures with the January release were numerical; they
looked to be the difference between the results of fpmath=387 (the
default for the x86 abi) vs fpmath=sse (the default for the amd64 abi).

That probably is still the case.

-JimC
-- 
James Cloos <[email protected]>         OpenPGP: 1024D/ED7DAEA6

_______________________________________________
Axiom-mail mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/axiom-mail

Reply via email to