This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: argument numbering for the tree inliner
- From: Mike Stump <mrs at apple dot com>
- To: Geoff Keating <geoffk at geoffk dot org>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Thu, 6 Nov 2003 18:07:10 -0800
- Subject: Re: argument numbering for the tree inliner
On Thursday, November 6, 2003, at 04:21 PM, Geoff Keating wrote:
You've changed the structure of the testcase for no apparent reason.
Please don't do that. It should be sufficient to just change the text
of the warning.
Actually, I've restored the structure of the test case to be more like
what it was. It was altered to be in its current form for no good
reason. The testcase is one I put in. The compiler ceased giving
warnings on statics, which motivated r1.2.
The testcase was supposed to be trimmed, and I failed to trim one part
of it, that if I had, it would have continued to work.
Here is my proposed change from the original test case:
bash-2.05a$ cvs diff -r 1.1 warn-1.c
Index: warn-1.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/gcc.dg/warn-1.c,v
retrieving revision 1.1
diff -u -p -r1.1 warn-1.c
--- warn-1.c 26 Oct 2002 01:15:16 -0000 1.1
+++ warn-1.c 7 Nov 2003 02:00:39 -0000
@@ -5,10 +5,10 @@
static void foo (p)
int p;
-{ /* { dg-warning "passing arg of" } */
+{ /* { dg-warning "passing arg 1 of" } */
}
-static void bar (void)
+void bar (void)
{
void *vp;
And here is the gook that was added, which I didn't like:
bash-2.05a$ cvs diff -r 1.1 -r 1.2 warn-1.c
Index: warn-1.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/gcc.dg/warn-1.c,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -p -r1.1 -r1.2
--- warn-1.c 26 Oct 2002 01:15:16 -0000 1.1
+++ warn-1.c 17 Apr 2003 11:46:56 -0000 1.2
@@ -14,3 +14,5 @@ static void bar (void)
foo (vp); /* { dg-warning "" } */
}
+
+void (*tourist_guide[]) (void) = { &bar };
The trimmed form of the testcase is preferable to the other form, the
original static isn't relevant to what is being tested.
Ok?