Hello,

As you may or may not be aware, part of my GSoC work is to set up a Jenkins
continuous integration server (see my blog post on the subject for some more
info[1]). As I announced on IRC a few days ago, there is a "test" server up
now on:

http://72.14.182.119:8080/

Anonymous users can see the build history, including the output for each
test run. Registered users can run testa and generally manage Jenkins. If
you'd like an account, please send me a mail or contact me on IRC. The
current installation is a "test" one so we can play around it; I'll set up
the "live" Jenkins installation later this week (Ondřej and me will
reinstall the whole server to the latest Ubuntu). Currently things work as
it should be expected, though there are some small issues to iron out. Once
we have the basic functionality set up, the idea is to somehow have Jenkins
run all tests on every pull request (a sympy-bot upgrade, if you will).
Before we do that, we should decide on some settings of Jenkins:

1) How often should Jenkins run the tests? Currently, we only do it
manually, but it's possible to schedule Jenkins to run at specific times
(daily, twice a day..), or after every commit? I think running after every
commit is a bit excessive (as we need to run the test suite at least 6
times, py25-py27 x gmpy, and that's already about one hour). I think running
twice a day would be more than enough for the amount of activity we have
currently, just daily is probably fine. On the other hand, the Jenkins
developers recommend running after every change (as that's the point of
continuous integration).

2) How should we report the results? The Jenkins interface is confusing at
best and it isn't realistic to expect all developers to check it all the
time. A common solution to this is to have it announce failures on IRC
automatically (we could report successes too, but that's a lot spam for no
practical gain). I could also set it up to send mails somewhere, but that
too might be excessive. If we run after each commit, though, it ought to be
possible to mail the author of the offending commit automatically. For the
moment, though, I think reporting failures on IRC is enough.

3) More of a longer-term consideration, but it would be useful to have at
least one additional Windows and Mac client. If anyone has suggestions...

4) Anything else to talk about?


Finally, I'd just like to mention that we've already found two test
failures/bugs with Jenkins, so there is definitely value to using it.


[1] http://vperic.blogspot.com/2011/05/continuous-integration-and-sympy.html

-- 
Vladimir Perić

-- 
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/sympy?hl=en.

Reply via email to