https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
Iain Buclaw changed:
What|Removed |Added
URL||https://github.com/dlang/dm
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
--- Comment #7 from Iain Buclaw ---
>From http://www.dsource.org/projects/dmd/changeset/259
The first setting of `const` to the result variable looks fine, as that's
against the parameter in the `out(result)` function, the result cannot be
chan
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
--- Comment #6 from Iain Buclaw ---
https://github.com/dlang/dmd/blob/4661fec6b792dcb22d66776fcdbe62ecb59667d1/compiler/src/dmd/funcsem.d#L2420-L2421
Introduced by.
https://issues.dlang.org/show_bug.cgi?id=3390
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
--- Comment #5 from Iain Buclaw ---
Looking at where TREE_READONLY gets set in the front-end, it's likely
introduced by PR110514.
It does look as though the result variable incorrectly has `const` tagged
against it though.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
--- Comment #4 from Iain Buclaw ---
The D front-end doesn't set TREE_STATIC on the `__result` variable, it's
gcc/gimplify.cc at this location:
```
/* If a const aggregate variable is being initialized, then it
should never be a lose to promo
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
--- Comment #3 from Iain Buclaw ---
It's only in latter tree dump passes we see that in the "bad" code gen, the
`const struct __result` is in fact static data from gdc-12 onwards.
```
struct str (struct B & this)
{
static const struct __res
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
--- Comment #2 from Iain Buclaw ---
Both gdc-11 and gdc-12 seem to produce the same tree structures from initial
dump.
```
struct toString ()
{
return = {.length=1, .ptr="1"};
}
void __invariant1 (const struct B & this)
{
return;
}
st
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119139
Iain Buclaw changed:
What|Removed |Added
CC||ibuclaw at gcc dot gnu.org
--- Comment #1
