[TIP] coverage.py and appengine

Ivo Bellin Salarin ivo.bellinsalarin at gmail.com
Tue Feb 7 07:41:57 PST 2017

Sorry for having bothered you. The problem seems to be unrelated to
coverage.py, but rather in the way the client calls the server.

I will report my advances, if any.

Le mar. 7 févr. 2017 à 15:32, Ivo Bellin Salarin <
ivo.bellinsalarin at gmail.com> a écrit :

> Hi everybody,
> has someone on this list been using coverage.py to collect code coverage
> on appengine applications, recently?
> I have been successful in collecting it, in an intrusive manner, hacking
> sandbox.py and stubs.py (because, in such environment, an application
> hasn't the right to write to the disk). Intrusive also means "modifying the
> code of the running application in order to call CodeCoverage API". My
> modifications occasionally generate concurrency errors (because of a
> failing assert in Collector.stop(), lines 306 and 307). These are the
> reasons for which I am not satisfied about this way of collecting CC.
> I have also tried to collect the code coverage in a non intrusive way
> (read: setting the COVERAGE_PROCESS_START environment variable and
> preparing an ad-hoc pth file that gets loaded during python processes
> startup). But in this case I collect *nothing* (apart from 5 statements out
> of 6 in /etc/python/sitecustomize.py). Perhaps because each appengine
> request is being processed in a particular environment where environment
> variables have a special meaning (and as such, the calling process
> environment variables have vanished).
> So, before abandoning the quest, I am searching for ideas or previously
> successful experiences.
> Just in case, I try to get CC on Ubuntu 16.04 and MacOS, using python 2.7
> (a standard appengine environment).
> Many thanks in advance,
> Ivo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.idyll.org/pipermail/testing-in-python/attachments/20170207/53982995/attachment.htm>

More information about the testing-in-python mailing list