Summary: | sci-mathematics/octave-2.1.73-r2 fails tests | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Bo Ørsted Andresen (RETIRED) <zlin> |
Component: | Current packages | Assignee: | Gentoo Science Mathematics related packages <sci-mathematics> |
Status: | RESOLVED WONTFIX | ||
Severity: | normal | ||
Priority: | High | ||
Version: | 2006.1 | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
/var/log/portage/sci-mathematics:octave-2.1.73-r2:20070427-120129.log
/var/tmp/portage/sci-mathematics/octave-2.1.73-r2/work/octave-2.1.73/test/octave.log |
Description
Bo Ørsted Andresen (RETIRED)
![]() Created attachment 117407 [details]
/var/log/portage/sci-mathematics:octave-2.1.73-r2:20070427-120129.log
Hi, Thanks for your report. Could you please try and see if replacing -Os with -O2 fixes this?? Thanks, Markus (In reply to comment #2) > Thanks for your report. Could you please try and see if > replacing -Os with -O2 fixes this?? Sure I can. :) Unfortunately it fails exactly the same both with -O2 and without -ggdb. :/ Thanks for testing! Could you please have a look at octave.log in /var/tmp/portage/sci-mathematics/octave-2.1.73-r2/work/octave-2.1.73/test and post some of the relevant errors/failures in an abbreviated file (the whole log is probably too large to post in its full glory). Thanks, Markus Created attachment 119277 [details]
/var/tmp/portage/sci-mathematics/octave-2.1.73-r2/work/octave-2.1.73/test/octave.log
Since it isn't really that big I'm attaching all of it. I hope that's ok.
There's a problem with the routine comparing the octave output from the tests with the expected result since the tests actually do pass. E.g. if you look at your ouput you see that running octave.test/arith/beta-1.m yields ans = 1 which is the correct result according to octave.test/arith/arith.exp. I suspect that this is related to UTF8 since you have LANG="en_GB.utf8". Any chance you could try compiling octave after switching to plain en_GB/en_US? cheers, Markus (In reply to comment #6) > Any chance you could try compiling octave after switching to plain > en_GB/en_US? Unfortunately the result is exactly the same. Both with LANG=en_GB and with LANG=C. :/ I see a lot of 'child process exited abnormally' in that file. Not sure if that can be the cause. (In reply to comment #7) > I see a lot of 'child process exited abnormally' in that file. Not > sure if that can be the cause. > Those are fine since they are actually supposed to fail (note the PASS after these types of tests). The problems are the ones with "ans=foo" which the analysis scripts fails to detect as PASS. I'll have a closer look at it. Thanks, Markus This version is gone from the tree |