Re: [PATCH 0/4] bitpos expansion summary reloaded

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

On Thu, 25 Oct 2012 18:51:23 +0200, Siddhesh Poyarekar wrote:
> On Thu, 25 Oct 2012 17:54:12 +0200, Jan wrote:
> > we will go with the fix without the full -Wconversion fix.
> > 
> > As most of the patches are very mechanical could you write a script
> > which will try to separate the "clear" and human-review parts?
> > 
> > "clear" I find that the patch is understood by the script (that is it
> > is some s/int/LONGEST/ etc.) and that it does remove >= 1
> > -Wconversion warnings.
> > 
> > Otherwise I will write such script.  I hope it would "approve" most
> > of the patches.
> 
> It would be great if you could do this.  I won't be able to get on it
> this week.

I have created the scripts below so far:
	http://people.redhat.com/jkratoch/conversiontest.pl
	http://people.redhat.com/jkratoch/buffer.c
	http://people.redhat.com/jkratoch/conversionsplit.pl

But it will be still several weeks of work to split and verify all the patch
chunks and after it is all done GDB will be still buggy wrt the 64-bit offsets
safety as was shown below.

Therefore going to drop it and start just normally fixing all the warnings for
-Wconversion -Wno-sign-conversion where the number of warnings is several
times larger but each such warning is simple and clear to fix.  Unless someone
has a better idea.


Regards,
Jan