PR 41469 followup: improve __builtin_stack_restore removal

[Paolo Bonzini]

On 11/25/2009 11:02 PM, Dave Korn wrote:
>    I don't see any smiley, so I'll just mention that I think the second two of
> those three items are really stretching a point.  There's a discussion over on
> the main list right now about exactly how far "obvious" goes, and I don't
> think anyone's suggesting it covers serious functional changes; and since part
> of the point of the approval process is for code review to be done, I'm not
> sure that approval should be that easily transferred to code that has not been
> explicitly reviewed and isn't even functionally identical (per your
> description of it as only "essentially" the same thing).
>
>    I'm not objecting to your patch, btw, just would like your input, since
> there's a thread about this topic going on right now.

I think it goes very much with the person.  Eric is an RTL maintainer, 
and probably nobody knows whether dwarf2out or parts of it is included 
or not.  Besides, his record of patches and especially reviews is such 
that very few people if anyone would care.

GCC rarely had any problems with "not-so-obvious" stuff checked in as 
obvious and when there were problems they were promptly solved by 
reverting or approving the patch.  In no case luckily we had to recourse 
to any kind of disciplinary action.  We should be proud of that :-) and 
use our judgement so that it stays that way.

Paolo