[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [RFC] Move buildbot master to ASF infrastructure

From: Bhuvaneswaran A <bhuvan_at_collab.net>
Date: Mon, 16 Nov 2009 17:29:47 +0530

On Mon, 2009-11-16 at 12:03 +0100, Lieven Govaerts wrote:
> On Mon, Nov 16, 2009 at 8:50 AM, Bhuvaneswaran A <bhuvan_at_collab.net> wrote:
> > On Sun, 2009-11-15 at 21:45 +0100, Lieven Govaerts wrote:
> >> > Given our familiarity with it, I think it makes sense to stick with
> >> > buildbot. That said, I like Hudson and we have done some work
> >> > recently so that our build and test can integrate nicely. For
> >> > example, Hudson can display the test results a lot more nicely and
> >> it
> >> > graphs things like the execution time that can help spot trends. I
> >> > personally prefer buildbots red/green model over Hudson's "weather
> >> > report".
> >> >
> >> The execution time of tests is (somewhat) visible in Buildbot's
> >> waterfall view, but I don't see that as a compelling feature. Are
> >> there any specific features in Hudson that you'd want that are not in
> >> Buildbot?
> >
> > We've the Hudson setup for Subversion build process.
> Are the Collab.net Hudson reports visible somewhere?

I'm not referring to CollabNet Hudson reports. Hyrum have setup one for
public Subversion couple of weeks back. Right now, it's not accessible
due to h/w failure.

> > Hudson provides many reports that I'm not sure if buildbot provides:
> > a) Test time trend report
> > b) Plugin to check the python coding semantics using pylint
> > c) Plugins for code coverage, Clover, Cobertura. I'm not sure if it has
> > support for any code coverage tool for C/Python though
> Clover and Cobertura can only be used for Java code, so that limits
> their usefulness to JavaHL.

Yeah, possibly there could be few plugins to cover C/python codes. We
may have to dig further into this.

> > d) ... and many more plugins that might be useful for this build process
> In general, Buildbot can run any code analysis or coverage tool that
> exists, as long as it can output a report that can be sent back to the
> server.
> The lack of specific plugins might lead to lower quaility reports though.

Same with Hudson. The plugins add more flexibility, but it's not
limited. The Hudson can be integrated with any application that can
generate the test result in junit format.

While we are at it, do we have access to report on the project/build
system mapping? I mean, out of "n" projects, how many projects use
Buildbot, Continuum, Gump and Hudson? This may give us a clue on usage
pattern of build systems for Apache projects.

> > It is easier to manage jobs in Hudson while we have to create new builds
> > for every new releases.
> It's not clear to me what you mean by this. Are you referring to the
> Collab.net certified builds?

When we have to qualify builds before every release, say, one before we
release 1.6.7, we may create a new job based on a existing one quickly.
Alternatively, we may use just one job on 1.6.x branch that we can use
to qualify all 1.6.x releases.

The job management is very easy.

> > As Greg suggested, we should try both the build systems to leverage the
> > benefits, until we decide upon anyone. I shall co-ordinate the setup for
> > Hudson, if we decide to do so.
> >
> Well, I don't know Hudson well enough to see the added value of Hudson
> over Buildbot, but if you're willing to do the work with ASF infra I
> don't see a reason not to try it.

Yeah, that's my point.

I wish you understand I'm not advocating Hudson over Buildbot. I'm
stressing, we try both the build systems in parallel for few releases
and stick to one (if need be) over a period of it, depending on how
comfortable it is for the community.

Bhuvaneswaran A    
CollabNet Software P Ltd.  |  www.collab.net

Received on 2009-11-16 13:00:28 CET

This is an archived mail posted to the Subversion Dev mailing list.