Re: Regression in exec.c (print_section_info) ?

[Joel Brobecker <brobecker at adacore dot com>]

>  > I would also like to see a new test if we don't already have one to
>  > prevent this type of regression in the future. 
> 
> It's easy to create a test for this specific case but I think it's
> very hard to test the exact format of all CLI output in general, and
> probably not worth the effort.  The next regression will likely occur
> elsewhere.

I disagree.

We already do a lot of testing of CLI output, and even if we didn't,
we have to start somewhere. "The next regression will likely occur
elsewhere" might be true, but that doesn't rule out the possibility
of regressing again. Since it is easy to create a test, why not just
do it, even if you think it's not going to bring actual gain? I like
to think of it as insurance. I hope that it will prove useless, but
it's still there if I need it.

> I only have one architecture: my old PC.

That's perfect.

> I didn't really think of it as a patch, I was just trying to draw
> attention to the regression.

This approach is too casual for me. Everyone else writes a ChangeLog,
which takes at most 1 min of your time. Please follow this part of
the procedure, so that I can review the ChangeLog entry as well.

> Actually it was a context diff - so you can read them!  They just look
> similar when lines are added.

Argh. Consider sending unified diffs, I think others prefer them too.
When I see context diffs, I usually end up converting it to unified...

Anyway, I ended up checking the attached:

2008-05-01  Nick Roberts  <nickrob@snap.net.nz>

        * exec.c (print_section_info): Add missing '\n'.

2008-05-01  Joel Brobecker  <brobecker@adacore.com>

        * gdb.base/info-target.exp: New testcase.

Tested on x86-linux. The new testcase fails without the change to exec.c.

-- 
Joel

Attachment: exec.c.diff
Description: Text document

Attachment: info-target.exp
Description: Text document