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

Re: Thoughts about src-dep implementation

Joey Hess <joey@kitenet.net> writes:

> debian/rules clean is already required to reverse the effects of the build.

In theory, yes.  But unless you audit the .diff.gz, it's not
necessarily obvious if this fails.  Now, I *do* audit my diffs, so I
know *exactly* where I've come up short (one package, which I very
recently adopted, where I did a mostly unedited upload just to get my
name in the maintainer field).  If I didn't audit my diffs, I think
I'd have at *least* four or five packages where the clean target
didn't do its job completely.

> Anyone who builds a package twice in the same directory knows this. As
> you've mentioned below, you don't. I expect most people do, though.

No, I suspect that most people simply don't notice, since the only
result is that the diff ends up a little larger than it would be

Of course, if your upstream maintainer is on top of things, and "make
distclean" works, then you probably don't have a problem.  But a lot
of times, that's not the case, and I've had to add extra cleanup code
to several packages.

The fact is that at the moment, if clean does a less-than-thorough
job, then garbage ends up in the diffs, but everything works fine, and
nobody notices.  However, if we start to depend on clean doing a
thorough job, I think we're going to have some unpleasant surprises.
Perhaps I'm overstating the pervasiveness of the problem, but it's
still a very real problem.

For an obvious example: *EVERY* gnome package I've looked at (balsa,
gnome-objc, gnome-libs, orbit) fails to clean up properly.  How many
of you knew that?  :-)

> My experience is counter to yours

Is it?  Or are you just not looking that closely?  I'm willing to
believe you only because I know you're a God of Packaging.  I suspect
that most of us mere mortals are not quite so careful.  :-)

Chris Waters   xtifr@dsp.net | I have a truly elegant proof of the
      or    xtifr@debian.org | above, but it is too long to fit into
http://www.dsp.net/xtifr     | this .signature file.

Reply to: