[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Is Sid for broken stuff? Is it too much to ask for testing the packages?

On Thu, Dec 12, 2002 at 01:56:05PM +0100, Marek Habersack wrote:

> > catch this bug is trivial, it is not always obvious to a maintainer when
> > more testing is needed, or *what* testing is needed -- or what changes
> Come on (I will stick to the postgres example) - if I add a test for a
> missing postgresql.env file then it's fairly obvious I have to remove that
> file and test the code. If the package breaks after installing it because of
> a programmatic error in the software itself, fine, we will cope with it.
> But, for all saint's sake, may the package install at least! Every single
> maintainer should test any change to their packaging scripts. Every, even
> the smallest change. And lack of time is not an excuse here because you
> rarely change large chunks of code in those scripts. If the test case is
> unclear then I agree with you, but cases like that missing postgresql.env or
> mozilla incorrectly calling an external script are painfully obvious --
> those packages should never hit even experimental.

The bargain of Debian unstable is this: the maintainer tests the package
to certain minimum standards, by ensuring that the package
can be installed, used, and (hopefully) uninstalled; and he also does
whatever other testing he feels he can.  In turn, the users get to use
the unstable packages, with the understanding that although the
maintainer has done his best, they may still contain bugs -- even "silly"
bugs -- and they may need to file bug reports to get the package fixed.
Any user of unstable who tries to make demands of developers beyond this
needs a reality check, because unstable is only viable so long as this
symbiotic relationship is present.

I'm happy that you have a strong work ethic when it comes to your
packages.  This does not mean it's reasonable to publically shame other
developers who does not live up to your standards, particularly when you
admit that you do NOT maintain such large packages, and you are NOT
personally willing to help with their maintenance.  When a maintainer has
little time, *thoroughly testing the package before upload is not
necessarily in the best interest of the users*.  Unstable exists because
developers are human, they make mistakes, they don't all have the same
background to know how to best test every change they make.  Their
fallibility does not mean their contributions should be belittled.  Being
a chronic pessimist, I'm sure there are some maintainers whose work is so
consistently poor that the project would be better off without them; but
unless you're willing to adopt their packages or request that they be
removed from the archive, dragging a developer through the debian-devel
mud is not going to improve anything.

Steve Langasek
postmodern programmer

Attachment: pgpWeww123UQ5.pgp
Description: PGP signature

Reply to: