Delay between cvs commit and build-request/build

[Elliot Lee]

The key assumption is "most packagers request build directly after
committing". That can be changed if the pre-build review needs it, e.g. to 
enforce approval.

Certainly more review is better. I believe that most of the package review
should be done after builds but before the package makes it into the main
repo. This is important because it allows the reviewer to look not only at
the .spec file but at the resulting packages. Frequently there will be
problems that aren't easy to find in the .spec file. Doing QA on the final
result means that we'll be looking at things from the same perspective as
the user: does the package install, upgrade, and run correctly?

Best,
-- Elliot

On Sun, 8 May 2005, Thorsten Leemhuis wrote:

> Hi *
> 
> We had something like the following once before, I can't find it in the
> archives (did not search very long), but it is somewhere. 
> 
> The purpose of the "Fedora Extras CVS commits <fedora-extras-
> commits at redhat.com>" Mailinglist was/is to make it possible for others
> like sponsors, package reviewers or other interested parties to check
> the fixes/changes that were committed to cvs before they get
> build/published. With the build-system now working automatically most
> packagers (including myself ;-) ) request build directly after
> committing. So actually no real check before build/publish (the latter
> actually depends on sign-time) is possible.
> 
> Should there be a delay in this process somewhere? At least 24 hours for
> example? 
> 
> Requesting build after commit is imho understandable and okay (if the
> publish can be canceled somehow) -- with this way you don't have to come
> back later to request the build (otherwise it might be forgotten if you
> touch a lot of different packages on one day). But maybe the packages
> should not be signed/pushed in the 24 hours (or maybe 48 hours) after
> the build-request/cvs-commit?!
> 
> Opinions?
> 
>