Testing of packages

[Michael Schwendt]

On Fri, 27 Apr 2012 12:10:11 +0200, MR (Matthias) wrote:

> Hi,
> 
> just a half-baked idea, I'm carrying some time with me around.
> 
> Some packages have a check-section being executed during build. Would it
> make sense to execute that test, when required packages of that package
> get updated (to see, if the update breaks my package)?
> 
> (Or other way around: when submitting an update run checks of dependent
> packages)

That depends on what tests are performed, how much a test-suite relies on
external interfaces, and what unexpected ABI/API/CLI breakage you hope to
catch that way.

I'd say, every test may be helpful in finding breakage. At a minimum, if
an inter-package dependency gets updated, one ought to ensure that a
program still starts and isn't affected by dropped symbols, for instance.
It may be more productive to skim over the changes in the component that
changed and conclude whether it's supposed to be a compatible update or
whether its changes ask for increased/longer run-time testing.
 
> If yes, how to automate this? Is there a way to access the check-section
> of a package from outside?

Hardly any %check section uses installed files. Typically, files in the
buildroot are used, or files plus test data in the builddir. Test data
won't be available.

-- 
Fedora release 17 (Beefy Miracle) - Linux 3.3.2-8.fc17.x86_64
loadavg: 0.06 0.05 0.05