Re: [patch] PR32901: handle assignment...(and PR34465 and PR34448)

[Diego Novillo <dnovillo at google dot com>]

On 12/19/07 07:44, Jakub Jelinek wrote:

> ATM non-empty CONSTRUCTOR on RHS (unless it is a vector CONSTRUCTOR) is
> considered invalid GIMPLE.

I said CONSTRUCTOR, but I really meant BIT_FIELD_REF.  However, I see 
your point and, yes, this is something we should combine at a later 
stage.  On our way out of SSA and GIMPLE looks like a good place.

Now, how much optimization are we losing here in real life?  I doubt 
these initializers are often used in hot paths.  So, I don't think 
there's a lot of urgency in fixing this.  But we should open a PR (if 
one doesn't exist yet).


> have some helper function which precomputes a lot of stuff into some temp
> structure, from which the predicate and gimplify_init_constructor then use it.

Yes, that's what I was thinking about.  This would mean stripping all 
the code out of the 'case RECORD_TYPE' handler.  Since we are pretty 
late in  stage 3, I guess this would be more surgery than we want at 
this point.

Would you guys be willing to re-arrange this for 4.4?  I'd rather do it 
now, but I understand if you want to wait.  In the meantime I guess this 
is OK.


Thanks.  Diego.