On Saturday, 9 November 2013 at 17:53:41 UTC, Johannes Pfau wrote:
Am Sat, 9 Nov 2013 17:01:43 +0000
schrieb Iain Buclaw <ibuc...@ubuntu.com>:
On 9 November 2013 12:19, Johannes Pfau <nos...@example.com>
wrote:
> The fix is simple:
> in gcc/convert.c(convert_to_integer):
>
> /* Convert to an unsigned integer of the correct width
> first, and
> from there widen/truncate to the required type. Some targets
> support the coexistence of multiple valid pointer sizes, so
> fetch
> the one we need from the type. */
> expr = fold_build1 (CONVERT_EXPR,
> lang_hooks.types.type_for_size
> (TYPE_PRECISION (intype), 0),
> expr);
>
>
That code does seem to contradict what the comment is saying.
I'll
raise it in #gcc and send a patch (unless you want to do it).
It indeed seems like it's only a small oversight. It'd be
great if you could report it.
Apparently it's working as per C semantics, which means the
behavior is undefined. I've mentioned that it makes no sense
that documentation/code don't match comments. But all I've heard
back is that you shouldn't use convert_to_integer in the front
end (except maybe through convert() - which I am 90% certain is
the case with us), and to implement our own semantics if we say
that pointer to integer conversion is clearly defined in D.
Regards
Iain.
