Re: [PATCH] Fix PR target/19933 and PR target/21315

[Mark Mitchell <mark at codesourcery dot com>]

Eric Botcazou wrote:

> > The bypasses look overly specific to me (presuming the precise form of a
> > future header will mention __GNUC__ on the line before a particular
> > definition).  As they are all for the benefit of a hypothetical future
> > header, I'd rather they just all bypassed on __GNUC__ and then if a future
> > Solaris version does get released with some but not all cases fixed then
> > the testcases you add should detect this and the bypasses can be refined
> > then.
>
>
> The reasoning is that we keep fixing unless we are sure we don't need to.  
> Using only __GNUC__ seems hazardous, as it could pop up in another unrelated 
> part of the header.

I agree with Joseph that this may not be the philosophy we want anymore.

In fact, I'd go so far as to say that maybe we should have a list of 
things that need to be fixed, not things that don't, on most platforms. 
 We've had at least two customers ask us to look through the fixes 
being applied to system headers, and when we've looked we've found a lot 
of things being fixed unnecessarily.  The unnecessary fixes have a cost 
in that they increase the problems with upgrading the OS, but not the 
compilers; you can rerurn fixincludes, but that's not trivial.

--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304