Re: [PATCH -mm] sys_semctl gcc 4.1 warning fix

[Al Viro <viro@ftp.linux.org.uk>]

On Wed, May 10, 2006 at 09:37:18AM -0700, Daniel Walker wrote:
> On Wed, 2006-05-10 at 17:21 +0100, Al Viro wrote:
> 
> > Don't.  Fix.  Correct.  Code.
> > 
> > Ever.  Because sooner or later you will paper over real bug.  It's far better
> > to reject patches that just make $TOOL to STFU than risk blind "fix" hiding
> > a real bug.
> 
> Couldn't agree with you more .. But I don't want to see the warning
> either ..

*shrug*
I hope that raw number of warnings is not used as job performance metrics.
All I can suggest is (a) watch for _changes_ in the warnings between revisions
and (b) get gcc folks fix the warning generation...

> > Unless you show a real codepath that leads to use without initialization
> > (and do that in commit message, so it could be verified as real issue),
> > these patches are worthless in the best case and dangerous in the worst
> > one.
> 
> Several of my patches have nothing to do with initialization .. 

s/codepath.*/bug being fixed/