Re: [PATCH] PR 23205

[Devang Patel <dpatel at apple dot com>]

On Oct 4, 2005, at 6:34 AM, Geoff Keating wrote:

> Hi Devang,
>
>
> > PR Debug/23205
> > * dbxout.c (dbxout_global_decl): Do not check DECL_RTL_SET_P.
> > (dbxout_symbol): Check DECL_RTL_SET_P, after handling constants.
> > Check NAMESPACE_DECL context for constants.
> >
> > * g++.dg/other/PR23205.C: New.
>
>
>
> > *************** dbxout_symbol (tree decl, int local ATTR
> > *** 2637,2644 ****
> > --- 2635,2646 ----
> >             HOST_WIDE_INT ival = TREE_INT_CST_LOW (DECL_INITIAL  
> > (decl));
> >
> >             dbxout_begin_complex_stabs ();
> > +           dbxout_symbol_name (decl, NULL, 'c');
> > +           stabstr_S ("=i");
> > +           /*
> >             stabstr_I (DECL_NAME (decl));
> >             stabstr_S (":c=i");
> > +           */
> >             stabstr_D (ival);
> >             dbxout_finish_complex_stabs (0, N_LSYM, 0, 0, 0);
> >             DBXOUT_DECR_NESTING;
>
> Don't comment out code; if it's no longer needed, just delete it.   
> Was this change intentional?  If so, what effect does it have?

I should have deleted it. dbxout_symbol_name() is a replacement for  
the commented name. It is preferable because it handles  
NAMESPACE_DECL also. I'll delete it before installing this patch.
> > + /* { dg-do compile { target powerpc*-*-darwin* } } */
>
> The testcase should not be limited to just ppc-darwin.  It makes  
> sense on every platform that supports stabs, so it should run on  
> all of them (and the next line eliminates the platforms that don't  
> support stabs).

OK.

> Also, I think it would be better if the testcase actually checked  
> that 'j' was output, rather than just that some constant got output.

I think we have other test case to  check it.


> (It might be good to have a more distinctive name than 'j', too.)

I'll rename it as 'foobar'.

OK with these changes ?
Thanks
-
Devang