[Incubator] autobuild/autotest for OSGeo projects

[Sean Gillies]

On Mar 26, 2006, at 9:16 PM, Frank Warmerdam wrote:

> Folks,
>
> I would like to suggest that an automated build and smoke test be part
> of the list of "highly desired features" of an OSGeo project.
>
> By automated build and smoke tests, I mean that project source code
> should be automatically built from CVS/SVN at least nightly and then
> some sort of minimal automatic test be run against that version to
> verify that the current version is at least minimally operational.   
> Ideally
> such build and smoke tests could be done more often, even on  
> demand.  And
> ideally the automatic testing should be fairly extensive.   
> Furthermore,
> ideally automatic notifications to developers, the developer  
> mailing list
> or somewhere similar should be done when the build or tests fail.
>
> Howard Butler has done a bunch of work with "Build Bot" to make this
> happen for GDAL and is now working on MapServer.  You can see an  
> example at:
>
>   http://gdal.builds.hobu.net/
>
> I am sure that some of the projects already do this.  I think that  
> GeoTools
> has something like this that is somewhat Java oriented.  Is that right
> Jody/Chris?  I also suspect that MapGuide already has this, though  
> I don't
> know if it is aimed at the OS (open source) version of MG.   
> Projects like
> Mapbender, and MapBuilder that are client browser oriented might  
> not make
> sense to approach in the same way.  Arnulf, Cameron, can you  
> comment on
> whether there would be a meaningful way in which build and smoke  
> test could
> be applied automatically?
>
> My intention isn't to mandate a particular approach to a build and  
> smoke test,
> but rather to push this as a highly desirable aspect of an OSGeo  
> project.  I
> do think that OSGeo ought to provide support to help setup this up for
> projects that don't have their own way they would like to do this.   
> To this
> end Howard has indicated a willingness to setup BuildBot support on  
> an OSGeo
> server at telescience for this purpose.
>
> I would add, that I see this as an action in support of the OSGeo  
> goal "To
> ensure a high degree of quality in foundation projects in order to  
> build and
> preserve the foundation brand."
>
> I have added this as an agenda item for the meeting tomorrow.  If  
> we are
> agreeable on this, my hope is that we can start setting up an OSGeo  
> build
> and smoke test server soon at telescience.  I have cc:ed Howard and  
> also
> Norman (who is responsible for OSGeo use of telescience systems) to  
> this
> end.
>

Frank,

As you might guess, because I've been nagging MapServer about this  
for a long time, I like this proposal. It seems that here we finally  
start to diverge from the ASF incubator model, which makes no such  
technical requirements of any project. Question is, when we start to  
make technical requirements, where do we draw the line? Build tests  
are good, but we really need functional tests to guard against  
problems like we have had with the MapServer 4.8 release.

cheers,
Sean

---
Sean Gillies
http://zcologia.com