This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFC/PATCH] Clean up unused variables (and prepare for `-Wunused-variable' flag)
- From: Sergio Durigan Junior <sergiodj at redhat dot com>
- To: Pedro Alves <palves at redhat dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Tue, 24 Apr 2012 15:03:58 -0300
- Subject: Re: [RFC/PATCH] Clean up unused variables (and prepare for `-Wunused-variable' flag)
- References: <m3ty0at632.fsf@redhat.com> <4F968B4D.3050209@redhat.com>
On Tuesday, April 24 2012, Pedro Alves wrote:
> Hi SÃrgio,
>
> First off, thanks for doing this.
Thanks for reviewing.
> On 04/23/2012 11:51 PM, Sergio Durigan Junior wrote:
>
>> Hi,
>>
>> This patch is a followup of the discussion in:
>>
>> http://sourceware.org/ml/gdb/2012-04/msg00171.html>
>> First of all, I am sorry for the size of this patch, but I couldn't
>> think of a good way of splitting it, and also I thought it would be
>> useless since these changes are all logically related.
>
>
> It's not useless at all. This warning points at two classes of problems:
>
> - variables that are no longer necessary, and can be garbage collected.
> - variables that actually should be being used, but they're not due to
> some latent bug.
>
> I skimmed the patch, and noted several places, mostly in tdep code, where
> you end up removing more than the unsuspecting auxiliary and obviously-left-
> -behind-by-accident variable. Some of those removed bits could well be latent
> bugs. Some hunks seem to remove used variables and expand what they were
> initialized from at the used sites. What's up with that? Please give rationale
> for any change that requires more than idle brain power to understand.
> :-)
We already talked on IRC, but I'm replying this message for the sake of
keep the record. As I said, initially I thought that those
initialized-but-not-used cases were copy-and-paste mistakes. But thanks
for pointing that. Here's what I'm going to do:
- Split these cases by arch, and resubmit it so that the maintainers of
each arch will get a chance to review and see if the case is a latent
bug or not
- Split and commit the obvious bits, as approved by Tromey and you in
the other message in this thread.
>> I'd like to apply it, but I have a couple of questions before:
>>
>> a) How's the ChangeLog for this patch supposed to be? Can I make a
>> "generic" ChangeLog, saying something like `Remove unused variables from
>> files'?
>>
>> b) I'd like someone to take a look at the `observer.sh' change, please.
>
>
> Please sent it as a separate patch, along with a rationale.
> Also, the .c files under features/ are generated files. We'll need to fix
> the generator instead, again, best done as a separate patch.
Thanks, I totally forgot about this fact! And I can't believe I didn't
read the header comment on the top of those files :-). Anyway, this
change will go as a separate patch as well.
--
Sergio