Re: Ideas for additional large scale tests
On Thu, Mar 01, 2007 at 02:59:59PM +0100, Stefano Zacchiroli wrote:
> I'm more worried about how you choose the environment you want to
> dist-upgrade. Various criteria that come to my mind (braindump):
> - base system only (pro: easy to set up, cons: tests a too small set of
> package)
> - \forall task, base system + 1 task (pro: still easy to set up, tests a
> lot more packages, I guess all tasks taken together are a fair share
> of the packages in the archive, cons: do not upgrade issues induced by
> inter-task relationships)
> - base system + a set of random selected packages (pro: easy to set up,
> tests inter-task issues, cons: non-idempotent, i.e. not easy to
> reproduce, no guarantees/idea about how much of the testing domain has
> been effectively tested)
- base system + all optional packages (which under policy are supposed to be
co-installable). This isn't going to give a small set of package
removals, but if done repeatedly over time one thing it allows is to
review each delta in the list of package removals *as it happens*, and
evaluate whether it's reasonable, rather than trying to fix these problems
all at the end of the release cycle.
Cheers,
--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
vorlon@debian.org http://www.debian.org/
Reply to: