Re: Another issue with diagnostic format-checker

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

"Zack Weinberg" <zack@codesourcery.com> writes:

| Gabriel Dos Reis <gdr@integrable-solutions.net> writes:
| 
| > "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu> writes:
| >
| > | I agree inclusion order dependencies is an issue in need of solving.
| > | Though perhaps "disaster" and "nightmare" is a bit overwrought. :-)
| >
| > I apologize if I sounded a bit rude, but changing something unsafe
| > (xxx_with_decl) to something safe (www + %D) and having that reject
| > sent me in agony-land.
| 
| Describing xxx_with_decl as "unsafe" is a bit overblown, don't you
| think?

No. 

| It has worked fine for years.

It APPREARS to have worked.  I should have pointed out that in the
patch 

  http://gcc.gnu.org/ml/gcc-patches/2003-07/msg01955.html

warning_with_decl was being passed a HOST_WIDE_INT
(e.g. larger_than_size) mistakenly described as an int ("%d".)

Kaveh himself has caught other unsafeties.

|  I am completely in favor of
| getting rid of it, on cleanliness and flexibility grounds, but I don't
| see where the code was unsafe.

Because you never actually tried to closely follow how it was used and
how it worked?

If the vairous functions were safe in the first place, there would not
be any point ini trying to put the checking-shield. Would it?

| > <front-end>/Make-lang.in is a good place, I would say.  What about:
| >
| >    BOOT_CFLAGS = $(BOOT_CFLAGS) -DGCC_DIAG_STYLE=__gcc_xxx__
| 
| This does not belong in the Makefiles.  Period.

I love this kind of position.

-- Gaby