Re: (gdb-6.8) Discard breakpoint address if shared library is unloaded

[Joel Brobecker <brobecker@adacore.com>]

Nick,

> You might find it disappointing that I've not rerun the testsuite but
> I find it disappointing that I have had to submit complete patches for
> code, doumentation and testsuite to revert changes which I pointed out
> broke things for Emacs.

Here is my point of view. It's my personal opinion on it and others
may have a different view on this, but FWIW: IMO, the reason why
a change "broke" emacs is because the GUD mode in emacs made some
undocumented assumptions on the format of that ouput of one of
the CLI commands. I recognize the importance of staying compatible,
particularly in a case like this where it seems relatively easy
to achieve. But you have to understand that nowhere did we promise
that we would make the CLI output "compatible". But again, we'll
do our best.

I have been more than willing to work with you on this, by delaying
the release and reviewing your patches, but I think we'll save each
other a lot of suffering if you accept that you are now in charge
of fixing this issue without trying to determine whether it should
be you or someone else that should be doing this.

So how about you submit a complete patch, following the usual protocol
rather than an abbreviated one? I'll review it promptly, and we can
put this episode behind us. As I said before, I reviewed the breakpoint.c
patch and it looked fine, so it's just a matter of taking a look at
the testsuite failures and adjust the testsuite accordingly (if
justified).

-- 
Joel