Re: [C++ Patch] PR 44516

[Paolo Carlini <paolo dot carlini at oracle dot com>]

Hi,

On 05/14/2012 07:38 PM, Jason Merrill wrote:
> On 05/14/2012 12:53 PM, Paolo Carlini wrote:
> > Yeah. I'm trying to replace *all* the current ones with the _loc
> > variant. The patch as you can imagine is becoming *huge*. Some details
> > will probably need another pass.
>
> One approach would be to make the non-_loc names macros that just call 
> the _loc variants with UNKNOWN_LOCATION.
But wouldn't this in turn require variadic macros which are C99-only? Or 
you mean something else?

Anyway, for this *specific* PR, if you agree I would rather follow your 
original advice and simply add build_min_nt_loc, and use it everywhere, 
thus removing completely the non-_loc variant (*) In a couple of 
situations I'm simply using UNKNOWN_LOCATION to be safe, this allows to 
replace completely the non-_loc variant and only two existing testcases 
must be adjusted, pure improvements (besides what appears in the 
testsuite, we are for example handling much better ARRAY_REFs and more).

Tested x86_64-linux.

Thanks!
Paolo.

(*) As you know, I tried to do more at once, but while I can easily see 
improvements in the testsuite, I can also see "random" regressions. I'd 
like to proceed incrementally.

/////////////////////////////////

Attachment: CL_44516_15
Description: Text document

Attachment: patch_44516_15
Description: Text document