[m-rev.] for review: deconstruct.named_arg for java

[Ian MacLarty]

On Wed, Jul 14, 2010 at 3:55 PM, Zoltan Somogyi <zs at csse.unimelb.edu.au> wrote:
> On 14-Jul-2010, Ian MacLarty <maclarty at csse.unimelb.edu.au> wrote:
>> >> That doesn't appear to solve the problem.  The function is actually
>> >> used.  What is needed is a "defined" attribute to convince gcc that
>> >> the function is actually defined.
>> >
>> > Since the state of things as they were before all this started actually
>> > worked, I think gcc was sufficiently convinced already.
>> >
>>
>> I'm not sure what you mean.  As far as I know gcc 4.x has always
>> emitted these warnings for the code generated in asm_fast.gc.
>
> Yes, but it generated code that assumed that the function was defined,
> and defined statically, which is all we need for correctness. The warnings
> are another issue. The fact that they are still there even if we give the
> unused attribute is a bit strange, since it seems to contradict the documented
> meaning of the attribute.

The documentation says "GCC will not produce a warning for this
function", but maybe it's meant to say "GCC will not warn that the
function is unused".  With the attribute it still seems to produce
warnings that the function is undefined. It's also possible I got the
attribute syntax wrong and gcc is ignoring the attribute.

Ian.

--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------