[Qgis-developer] Rethinking the testing and release procedure of QGIS

[George Rodrigues da Cunha Silva]

On quinta-feira, 30 de junho de 2011 16:55:09, Andreas Neumann wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi all,
> 
> Please regard my email as constructive criticism. This is as much a
> criticism of the users (including myself) as it is on the developers. It
> is my impression that the users are not testing QGIS trunk enough before
> the release, but they complain after the release.
> 
> I am having serious concerns with the quality of the final QGIS 1.7
> release as it currently stands. I have to say that I am usually using
> trunk. The trunk version, it seems to me, has fewer issues than the
> release versions. Also it seems, that the self-compiled trunk version
> works fine, but the Ubuntu packages may have problems. Then there are
> the PDF/printing issues on Windows ... which are major problems to many
> people.
> 
> I think we should, in the future, first release a beta-version and test
> it for a couple of days/weeks before we call it a final rele
ase. Also,
> we should do - much more - testing. Testing is the task of the power
> users, not so much the developers. And we need testers, not only on
> Linux, but also on Windows and Mac. While many devs are on Linux, the
> majority of the QGIS users work on Windows.
> 
> Maybe the establishment of a testing infrastructure, as proposed by Strk
> and others, would also help to maintain/raise the quality.
> 
> Another idea is to collect money (something like a membership fee
> (voluntary, but expected from users who use QGIS a lot)) and pay QA
> people. With the fast growing QGIS user base it should be possible to
> collect enough money for such a task.
> 
> It is my impression that the current state of the 1.7 version with all
> the major issues hurts the reputation of QGIS. We need to act and make
> sure that this does not happen again. The QGIS team has to become more
> professional if it wants to grow its user base and doesn't want to loose
> their users. This includes bo
ring tasks, such as QA.
> 
> Any thoughts?
> 
> Andreas
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.11 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
> 
> iQEcBAEBAgAGBQJODNSaAAoJELiCsGDopvBCT+0IAIOYTjIAETPIa5Amu4Oa1yP5
> jM3HGSsqPpFVSWnFVuFAgLbziiMBJ8O+u6ucsTvoZWnxiHNLs5b+t5NIfDdtTulo
> TLOFCVQDVtv5mOBe7CJpVpj9POToP+VpAZqeg5ni4OZ28JbQzWLyV6u9AJCUFwnq
> k3gcClZ2FWw31la7ABcTc8LT+JJv7zpQCJ7M3O5apatSg54xiVQSxIb5mlzNOtwg
> 9JYsdMRHvxm2UB+6HXgIJ8BPhh6OTR5OWFOturwFeoTBPXdsUrRlkClZSgdNrTQy
> 3azBy/XcnkVVvo4r9EMfD4N3fQeENp/PzVKEGlP+pd6+4y8GWvZ63ZZr+4PvDvY=
> =VoH6
> -----END PGP SIGNATURE-----
> _______________________________________________
> Qgis-developer mailing list
> Qgis-developer at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/qgis-developer

I agree that QA testing should be done by power users. The problem is 
that testing software is exhausting and we could, perhaps, add a new 
tool to the many we use, something like Test
Link.

The developers would create the tests, and we could assign them to a 
few voluntaries? A tool like this can help cover the multiplicity of 
white, gray and black box tests we need.

George

-- 
--------------------------------
George Rodrigues da Cunha Silva
Desenvolvedor GIS
http://geoprocessamento.net
http://blog.geoprocessamento.net