Re: Debug symbols in common block

[Toon Moene <toon at moene dot indiv dot nluug dot nl>]

Richard Henderson wrote:

> On Tue, Feb 15, 2000 at 07:02:15PM -0700, Andy Vaught wrote:

> > Its almost the same patch as before, but one of the #ifdef/#endif's that
> > is around a section of code that handles equivalence statements has been
> > left in.  g77 doesn't crash on your test program and I still get symbols
> > in common blocks.

> Toon, if you can ack this not crashing for a reasonable
> set of inputs, I'll check it in.

I ran it through a set of 6 different inputs that I thought up
specifically to test this problem with and through our 1300+ routine NWP
code.  It passes with flying colours.

BTW:  The original solution even caused a f771 crash on:

      program aap
      equivalence (i,r)
      i = 2
      print*,r
      end

So my suggestion relating the crash on equivalence with dummy arguments
was a red herring.

Of course this means that this patch is only a solution for items in
common blocks, not in equivalences - we should not forget to emphasize
that in the documentation and the news item for 2.96.  Nevertheless:
very good work, Andy !

Cheers,

-- 
Toon Moene (toon@moene.indiv.nluug.nl)
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
Phone: +31 346 214290; Fax: +31 346 214286
GNU Fortran: http://gcc.gnu.org/onlinedocs/g77_news.html

PS: Richard, don't you forget to repair the typo in the following patch
    to the release branch:
*** libf2c/libI77/endfile.c.orig        Sun Jan 23 12:06:22 2000
--- libf2c/libI77/endfile.c     Sun Jan 23 21:10:55 2000
*************** t_runc(alist *a)
*** 104,107 ****
--- 104,108 ----
        if (copy(tf, loc, bf))
                goto bad1;
+         b->uwrt = 1;
        b->urw = 2;
  #ifdef NON_UNIX_STDIO

     You mistakenly wrote `b->urwt = 1;' there :-(