On Thu, Sep 12, 2013 at 2:29 AM, Roland Mainz <[email protected]> wrote:
> On Mon, Aug 19, 2013 at 6:31 PM, Glenn Fowler <[email protected]> wrote:
>> we will most likely handle "restrict" in a similar way to "const"
>> to keep the code readable
>> where
>> all ast code will use "restrict"
>> #ifdef magic in ast_common.h will do one of the following
>> possibly augmented by probe info when nmake is used to build:
>>         (1) nothing -- "restrict" or its expansion will be seen by the 
>> compiler
>>         (2) #define restrict __restrict
>>         (3) #define restrict
>> with (3) being the fallback position
>
> Ok...
> ... can we do this for the next alpha, please (technically... without
> consumers... it should not harm anything (currently) ... right ?) ?

Erm... ping!
... can we do this for the next alpha, please (just the infrastructure
to minimise the risk... and then make step-by-step patches to convert
the possible consumers to claim our 1.6-3.8% performance benefit (this
is the average the clang people measured across all software they
tested with+without |restrict|)) ?

----

Bye,
Roland

-- 
  __ .  . __
 (o.\ \/ /.o) [email protected]
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 3992797
 (;O/ \/ \O;)
_______________________________________________
ast-developers mailing list
[email protected]
http://lists.research.att.com/mailman/listinfo/ast-developers

Reply via email to