Re: Beyond GCC 3.0: Summing Up

[Alexandre Oliva <aoliva at redhat dot com>]

On Jul 10, 2001, Richard Earnshaw <rearnsha@arm.com> wrote:

> go and fix the problem you exposed (which may then throw up other
> problems)

Which is a very interesting point.  So far, we (or I :-) have only
considered the case of a single patch exposing a single latent bug.
Bug it's very well possible that the patch that fixes the latent bug
exposes yet another latent bug.

So, should we require the patch that exposed the first latent bug to
be reverted, its proponent would now have 3, 4 or more bugs to fix,
quite likely in totally different parts of the compiler.

By leaving the patch in so that others more knowledgeable in those
parts have an extra push to attempt to fix it, we'll be not only
sharing the burden among all interested parties, but also dealing with
potentially long chains of latent bugs without introducing dead-locks
in case the proponent doesn't have the expertise needed to fix a
problem, and the people who do don't see the problem because they
missed the patch.

It's an unfortunate truth that a large number of patches remain
unreviewed for long periods of time.  Why do we have reasons to
believe that patches that happen to expose latent bugs and then end up
reverted would get any more attention than patches that just remain
unreviewed?

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me