[Neuroimaging] Upcoming nipy release

Yaroslav Halchenko lists at onerussian.com
Fri Sep 25 02:30:33 CEST 2015


On Thu, 24 Sep 2015, Matthew Brett wrote:
> >> Our setup is https://github.com/datalad/buildbot which is in large based on
> >> https://github.com/ethereum/ethereum-buildbot and apparently work on supporting
> >> pull_requests in stock buildbot since then was accepted!
> >> https://github.com/buildbot/buildbot/pull/1632
> >> so I might look into redoing it using stock features

> > That would be really good - if you do have time to do it.  It would be
> > ideal to get the buildbots integrated into the github interface...

> After thinking about it for a little bit, I was worried about a)
> overwhelming the buildbot machines, and b) allowing any PR to execute
> code on the buildbot machines.

our setup restricts automagic testing of PRs only for a limited set of
github logins.  For the rest of PRs I (or other developers with enough
rights) would need to add a label 'buildbot' to the PR to trigger
buildbot considering it (so similar to homu's approvals)

> I wonder whether we could use Homu
> instead?
> https://github.com/barosl/homu

I thought I saw all of the possible solutions when I was setting it up a
while back, may be Homu too... there seemed to be quite a bit of new
development in it since then.

> I think this would mean that approved reviewers could mark the PR with
> a comment like '@homu-user r+' to the PR, and this could trigger
> builds both on travis-ci, and the buildbots, and the PR would only get
> merged if the all the tests pass.  I think.

> That seems like a good mix - an approved reviewer has to OK it before
> it is tested on the buildbots and travis before merging.

> What do you think?

yeah -- sounds good, but on the other hand requires additional
interaction (unless there are automatic "approvals" for testing for some
logins). it is for sure probably better than the ad-hoc solution I
adopted, but it might still be better to re-review what is the built-in
support within buildbot atm.  It might be easier to provide similar
logic if not present "natively"

-- 
Yaroslav O. Halchenko
Center for Open Neuroscience     http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik        


More information about the Neuroimaging mailing list