[Biopython-dev] Coverage metrics

[Peter Cock]

On Mon, May 25, 2015 at 3:57 PM, Tiago Rodrigues Antao <tra at popgen.net> wrote:
> On Mon, 25 May 2015 15:25:07 +0100
> Peter Cock <p.j.a.cock at googlemail.com> wrote:
>> By all means give that a go - the TravisCI configuration will
>> work on personal branches fine (that's how I've tried things
>> prior to pushing them to biopython/master).
>
>
> In terms of designing implementation, and to avoid excessive online
> usage I was planning on doing the following:
>
> Have two branches. The main one would do offline testing. There would
> be a second branch that would do online testing. Every week an
> automated process would push changes from the main to the online one.
> The .travis.yml code would check the branch name and take care of
> everything (to the online branch would be just an annotation, really).
>
> Does this sound a good idea?
>
> Tiago

That sounds overly complex with messy details like public
branches too visible.

I think I'd prefer a hack inside the TravisCI script which said
if this was the master branch (not a merge for a pull request
etc), and it was Sunday (for example), then run more tests
(e.g. online, more dependencies, with coverage).

[If there is more state information available within Travis,
this could make a more sensible choice like run the full
tests linked to when that was last run.]

This reminds me I've not yet resurrected an online PyPy
test target for the buildbot farm:
http://lists.open-bio.org/pipermail/biopython-dev/2015-May/020931.html

Peter