This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Delete EXPR_LOCUS
2009/4/24 Joseph S. Myers <email@example.com>:
> On Thu, 23 Apr 2009, Eric Botcazou wrote:
>> this patch removes EXPR_LOCUS and all traces of indirection from the source
>> location business. ?This means changing the signature of the 'H' specifier
>> from location_t* to location_t, hence a few warnings during stage #1.
> If you're changing every call using %H (or many such calls), making them
> use warning_at, error_at etc. (with the location as an explicit prototyped
> argument) might be better. ?I think we'd like eventually for all calls to
> diagnostic functions to pass an explicit location argument (and maybe
> eventually for error_at to be renamed to error etc.).
I am updating and testing two patches to do precisely this for warning
and error. Immediately after, %H will be removed. So using warning_at
and error_at will facilitate me things. Besides, %H is unreliable in
case of system headers. I think the other location specifiers (%K, %J)
should also be removed but I am not sure how trivial that will be.