-
-
Notifications
You must be signed in to change notification settings - Fork 18.4k
few fresh failed tests on big-endian platforms (0.17.0+git8-gcac4ad2) #11282
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
actually time_series ones aren't BE specific since happen also on armel: https://buildd.debian.org/status/fetch.php?pkg=pandas&arch=armel&ver=0.17.0%2Bgit8-gcac4ad2-1&stamp=1444493597 |
pls |
once again -- those failures are architecture specific -- pass fine on conventional x86 (32 or 64 bit). Those full reports above always contain the printout of show_versions like this one on armel:
|
do you have the chart of how many users actually use these platforms? would be interested to see. |
if overall, not just for pandas, then see http://popcon.debian.org/ . Remember that this is voluntary submissions, actual number of users is somewhere between x10 and x100 of that ;) Also note that failures on big endian platforms are quite often not big endian specific but rather trigger bugs in the code which otherwise go not mentioned on regular little endian, e.g. that one |
nice graphs! ok so after i386/amd64 things drop off a lot. |
sure thing, but e.g. more and more arms you might see around, so who knows how it will look in e.g. 10 years? ;) |
some of those persisted in 0.17.1: https://buildd.debian.org/status/fetch.php?pkg=pandas&arch=s390x&ver=0.17.1-1&stamp=1449119537 |
I confirm that for 0.17.1 under s390x we have this:
|
On Ubuntu for arm64 for 0.17.1 we're hitting:
|
Partially fix pandas-dev#11282 for s390x
Also on Ubuntu for powerpc for 0.17.1 we're hitting:
|
love to have fixes for these. but keep in mind that we don't formally support these platforms as they are not regularly tested. so fixes would have to be specifically targeted with tests (see my comments on #12269). |
note that NaT issue is deep within numpy guts: numpy/numpy#8325 |
@yarikoptic can you confirm whether these failures persist with up-to-date versions? The constraint on trying to troubleshoot these architectures is that we don't have them available in the CI. I don't suppose Debian has a "Guest CI" program for exactly this situation? |
While we do want to fix errors on various platforms the biggest challenge is finding a CI providers that support various architectures. Additionally, these test examples appear to have run on fairly old versions of pandas. Due to these issues and the inactivity going to close. Happy to reopen if someone could confirm these issues persist on master |
for details see https://buildd.debian.org/status/fetch.php?pkg=pandas&arch=powerpc&ver=0.17.0%2Bgit8-gcac4ad2-1&stamp=1444485115 but here is the excerpt of few failed tests:
The text was updated successfully, but these errors were encountered: