[PATCH][4.5] Move gimplifier predicates

Andrew Haley aph@redhat.com
Fri Dec 5 11:16:00 GMT 2008


Richard Guenther wrote:
> On Thu, Dec 4, 2008 at 7:01 PM, Andrew Haley <aph@redhat.com> wrote:
>> Richard Guenther wrote:
>>> This moves gimplifier predicates to where they belong and makes them
>>> private.  Apart from the uses in walk_gimple_op for which I have no
>>> clue what this monster-function tries to do - Diego, do you remember?
>>>
>>> The java use seems unnecessary (as the comment hints).
>> I wrote that comment, and to my eyes it hints no such thing.
>>
>> As the comment says, it was intended to make well-defined an expression
>> such as
>>
>>   a + ( a = 2; )
>>
>> by transforming it into
>>
>>   ({ A = a; A + ( a = 2 ) })
> 
> The comment says " Or perhaps that's unnecessary because
> the front end already generates SAVE_EXPRs." - or does that refer to
> the parameter list case only?

It refers to the parameter list case only.

> Anyway, the testsuite is clean with that patch - a testcase that breaks would
> be nice to have in the testsuite.

It is quite likely that the new java front-end no longer generates
expressions of this form.  I think temporaries are always produced.

Andrew.



More information about the Gcc-patches mailing list