On Fri, Feb 7, 2014 at 10:03 PM, Sriram Ramkrishna <[email protected]>
wrote:
The idea is to test everything wtih as much automation as possible
while finding interaction bugs in the manual test. Now that we have a
continuously integration
gnome-continuous is continuous *delivery*. This is actually quite
important - you can download and run the result.
(If you count the smoketest as a deployment, then it's continuous
deployment)
If a test fails, you really want the ability to download the *exact
same*
binaries that the build server is testing so you can use gdb locally.
This
is something that was always irritating to me with rpm/dpkg using
"make check" and jhbuild buildbot.
Phase 1:
* define the list of components that will be QA'd
This all sounds unnecessarily formal, plus...the list of components
that will be
QA'd would equal the list of components we ship, I hope =)
Personally, I would start with trying to spread
https://wiki.gnome.org/Initiatives/GnomeGoals/InstalledTests
both inside GNOME, to downstreams, and to other ecosystems.
Taking the tests we already had written and automating them,
running them *all of the time* has been a huge win.
_______________________________________________
[email protected]
https://mail.gnome.org/mailman/listinfo/release-team
Release-team lurker? Do NOT participate in discussions.
