This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Fix overzealous folding

On Fri, May 30, 2008 at 9:36 AM, Eric Botcazou <> wrote:
>> Eric, would you please look into this
> Already done, please re-read the thread.
>> and work out what we really ought to be doing here?
> Already done as well: properly disable the folding or rewrite VRP to avoid
> taking advantage of TYPE_MIN/MAX_VALUE.
>> And, would you please add a comment to the code that you changed that
>> indicates that the exclusion of ENUMERAL_TYPEs is a work-around, with a
>> link to the PR so that we know why?
> Here's what I've installed on the mainline.  The exclusion of BOOLEAN_TYPE has
> been there from day #1 apparently.

Yep.  Note that fold is really a special place in GCC - it is used from both
the frontends and the middle-end and seemingly arguing from middle-end
semantics point may well break the frontends if they (wrongly?) assume
that the (not written down) middle-end semantics do not apply, but their
frontend semantics.

I wouldn't be surprised if removing the BOOLEAN_TYPE check broke
C or C++ at least ;)


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]