OK, thanks.
On Tue, Dec 21, 2010 at 3:29 PM, Sun Chan wrote:
> Jiangzhou,
> yuor fix is good to go
> Sun
>
> On Tue, Dec 21, 2010 at 1:40 PM, Jiangzhou HE
> wrote:
> > SCLASS_FORMAL_REF refers to pass-by-reference formals in Fortran.
> >
> > Since SCLASS_FORMAL_REF can be modified in the same w
Jiangzhou,
yuor fix is good to go
Sun
On Tue, Dec 21, 2010 at 1:40 PM, Jiangzhou HE wrote:
> SCLASS_FORMAL_REF refers to pass-by-reference formals in Fortran.
>
> Since SCLASS_FORMAL_REF can be modified in the same way as SCLASS_AUTO and
> SCLASS_FORMAL, I think it is also necessary to add the gu
SCLASS_FORMAL_REF refers to pass-by-reference formals in Fortran.
Since SCLASS_FORMAL_REF can be modified in the same way as SCLASS_AUTO and
SCLASS_FORMAL, I think it is also necessary to add the guarded code for
SCLASS_FORMAL_REF.
Thanks.
On Tue, Dec 21, 2010 at 2:25 AM, Robert Kennedy
wrote:
Sun Chan said:
> you are right, I did not look careful enough.
I haven't looked at the surrounding code, but I agree that on the
surface the proposed change looks correct, at least for the
SCLASS_FORMAL case. I don't remember the semantics for
SCLASS_FORMAL_REF so I'm not sure about that part of t
> from the numbering, it seemed to be sgi bug numbers. Which means
> Robert Kennedy (the original author of this code) added the comment,
> for obvious reasons.
I think Sun is right; those are my comments. Unfortunately I don't
remember the details of those bugs nor the test cases they would have
you are right, I did not look careful enough.
Sun
2010/12/20 Jiangzhou HE :
> The current code uses ST_IDX_level(ST_st_idx(st)) == CURRENT_SYMTAB to
> determine whether an auto variable is uplevel ref. My fix follows the old
> way to do that. What is the right way? Thanks.
>
> On Mon, Dec 20, 2010
The current code uses ST_IDX_level(ST_st_idx(st)) == CURRENT_SYMTAB to
determine whether an auto variable is uplevel ref. My fix follows the old
way to do that. What is the right way? Thanks.
On Mon, Dec 20, 2010 at 7:46 PM, Sun Chan wrote:
> I thought there is a bit in the symbol table that tel
Yes, x is SCLASS_FORMAL. I meant the code treated x in the same way as a
local variable. By adding guarded code for SCLASS_FORMAL, x would be treated
in the same way as an uplevel auto variable.
Thanks.
On Mon, Dec 20, 2010 at 8:21 PM, Tianwei wrote:
>
>
> On Mon, Dec 20, 2010 at 7:25 PM, Jiang
from the numbering, it seemed to be sgi bug numbers. Which means
Robert Kennedy (the original author of this code) added the comment,
for obvious reasons. The code must have a good reason to be there.
Please read through the code again and make sure we are looking at
uplevel ref the right way
Sun
On Mon, Dec 20, 2010 at 7:25 PM, Jiangzhou HE wrote:
> Hi,
>
> Can a gatekeeper help to review the patch to bug 702 (
> https://bugs.open64.net/show_bug.cgi?id=702)
>
> This is the test case:
>
> #include
>
> void foo(int x) {
>
> void far() {
> x++;
> }
>
> void bar() {
> int x2 =
I thought there is a bit in the symbol table that tells whether the
symbol is uplevel ref. Your method seemed to be a hack
Sun
On Mon, Dec 20, 2010 at 7:25 PM, Jiangzhou HE wrote:
> Hi,
>
> Can a gatekeeper help to review the patch to bug 702
> (https://bugs.open64.net/show_bug.cgi?id=702)
>
> Th
Hi,
Can a gatekeeper help to review the patch to bug 702 (
https://bugs.open64.net/show_bug.cgi?id=702)
This is the test case:
#include
void foo(int x) {
void far() {
x++;
}
void bar() {
int x2 = x;
far();
printf("%d %d\n", x, x2);
}
bar();
}
int main() {
foo(1)
12 matches
Mail list logo
