Where are all our testing machines? #1153

Closed
jseabold opened this Issue Oct 25, 2013 · 8 comments

Projects

None yet

3 participants

@jseabold
Member

We get travis, pythonxy, nipy, ubuntu testing? Anything else? What are the URLs. If we collect them here, I'll make a PR adding a landing page for these to the docs.

@josef-pkt
Member

That's it, just the three.

https://travis-ci.org/statsmodels/statsmodels/builds
http://nipy.bic.berkeley.edu/waterfall?category=statsmodels
https://code.launchpad.net/~pythonxy/+recipe/statsmodels-daily-current (*)

plus coverage
https://coveralls.io/r/statsmodels/statsmodels

we don't have a home yet for the vbench, PR is waiting and ready

(*) which is currently red, which means a build failure not a test failure. test failures still leave it green.

All the other version compatibility testing from python 2.6 to 3.3, and various versions of our dependencies, happens on my computer, when I run my tox tests.

@jseabold
Member

How do we get the build log of the failure?

@josef-pkt
Member

I don't know

the Ubuntu version top line has a build log, but I don't see a problem in there, nor do I know if this is the right log
example https://launchpadlibrarian.net/154937999/buildlog.txt.gz

maybe an internal error because it stopped after 20 seconds.

@jseabold
Member

I can set up the vbench as part of the doc builds. We just need to update this at release right?

@josef-pkt
Member

We just need to update this at release right?

not sure what you mean.
We would like to run this in regular intervals, so we can get some feedback when we hurt or improve our performance.

@jseabold
Member

I just mean we don't need nightly vbench builds, so it's something I could do by hand when I upload release docs.

@jseabold
Member

I.e., you can run them by hand and check perf but we only need to advertise at release.

@TomAugspurger
Contributor

Yep, wouldn't need to be run nightly since it will go through the git history. And then before a potentially performance sensitive commit the submitter would run bench on the branch.

I may need to work on the test_perf part some more. It may have some issues on non-Linux systems. I can't even run the pandas vbench on my Mac right now.

@jseabold jseabold closed this Apr 2, 2014
@jseabold jseabold added a commit to jseabold/statsmodels that referenced this issue Apr 3, 2014
@jseabold jseabold DOC: Add links for build machines. Closes #1153 [skip ci] 40ed7a8
@PierreBdR PierreBdR pushed a commit to PierreBdR/statsmodels that referenced this issue Sep 2, 2014
@jseabold jseabold DOC: Add links for build machines. Closes #1153 [skip ci] 2ba56ef
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment