[Pbuilder-maint] pbuildd script in pbuilder alioth?

Daniel Schepler schepler at math.berkeley.edu
Wed Apr 5 08:44:41 UTC 2006


Le Mercredi 05 Avril 2006 02:29, Junichi Uekawa a écrit :
> My current policy with pbuilder is that it is regression tested once
> every day at 12:00 JST.  I don't know if it's feasible with pbuildd to
> do a regression testing, since I guess it will fail more often than
> not.
>
> Do you have anything in mind for QA work of pbuildd?

I hadn't thought about it -- currently every once in a while I move everything 
from pool/ into old-pool/ on my machine and start everything over, and test 
things that way.  I suppose it might also be useful to test everything in 
scripts/cb/ periodically to detect any breakages in the scripts themselves 
against updated source packages (as opposed to new Build-Depends loops I have 
to take care of, which I don't see any way of detecting automatically).

Currently scripts/cb-functions is completely broken since the recent upgrade 
to su, but I haven't had time yet to look at what I need to do to fix that.

On a related note: one of the reasons for the fragility of scripts/cb/ is that 
I currently have to list my custom Build-Depends in full.  I've thought that 
it would be nice to be able to instead list what differences I want from the 
package's Build-Depends (i.e. "ignore po4a", "replace libgl1-mesa-dev with 
xlibmesa-gl-dev", etc.)  I think the best way to implement this would be to 
get support from pb-satisfydepends to be able to pass a list of Build-Depends 
on the command line in place of a .dsc file.  (Another reason is that for 
some reason the places being patched tend to change extremely often, but 
there's probably nothing to be done about that.)
-- 
Daniel Schepler



More information about the Pbuilder-maint mailing list