TST/BUG: stata failures on big endian

[jreback]

related #5778

Investigate/fix whether the test failures (currently skipped on big-endian) are actual bugs or just an odd test.

cc @PKEuS

@jreback , what where the other failing tests? need to mark everything dirty until fixed.
I suspect the user base for stata on sparc numbers somewhere in the lower 0'z, so perhaps
it's just fine to blacklist the whole thing and call it duly mitigated.

If this isn't resolved by 0.14 release time please ping me and I'll just nuke it from space.

[jreback]

@y-p there are 5 skips; my brief investigation is that it needs to do some kind of byte swapping in order to compose some of the data on read-back. I don't know enough about what its supposed to do.

[yarikoptic]

Hi guys -- would anyone have time to look at this one?
It keeps buildbots unhappy on the sparc box (http://nipy.bic.berkeley.edu/builders/pandas-py2.x-sid-sparc/builds/768/steps/shell_7/logs/stdio) and other glorious architectures e.g. https://buildd.debian.org/status/fetch.php?pkg=pandas&arch=powerpc&ver=0.14.0~rc1%2Bgit79-g1fa5dd4-1&stamp=1401244093

Would anyone have time/interest to look into it?

[jreback]

just going to disable the tests (again)

[yarikoptic]

oh no... ok -- let me look into it at least briefly.

On Wed, 28 May 2014, jreback wrote:

just going to disable the tests (again)

—
Reply to this email directly or [1]view it on GitHub.

References

Visible links

1. TST/BUG: stata failures on big endian #5781 (comment)

Yaroslav O. Halchenko, Ph.D.
http://neuro.debian.net http://www.pymvpa.org http://www.fail2ban.org
Research Scientist, Psychological and Brain Sciences Dept.
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834 Fax: +1 (603) 646-1419
WWW: http://www.linkedin.com/in/yarik

[jreback]

I think the reading/writing is prob incorrect (e.g. may need byteswapping on big endian) as these are raw byte reads...but noone I know has a big endian :) so not sure what can do about it

[yarikoptic]

On Wed, 28 May 2014, jreback wrote:

I think the reading/writing is prob incorrect (e.g. may need byteswapping
on big endian) as these are raw byte reads...but noone I know has a big
endian :) so not sure what can do about it

I think I gave you access to a big endian already... at least I see a
jreback user on that vagus box ;)

Yaroslav O. Halchenko, Ph.D.
http://neuro.debian.net http://www.pymvpa.org http://www.fail2ban.org
Research Scientist, Psychological and Brain Sciences Dept.
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834 Fax: +1 (603) 646-1419
WWW: http://www.linkedin.com/in/yarik

[jreback]

I know, that's how I debugged the value_counts issue :)

no time to do anything about these stata issues (and its a low-level code reading)...this issue will stay open, maybe someone will get to it at some point