On Jun 16, 2005, at 1:26 PM, Roberto Bagnara wrote:
First, I would like to clarify I do not consider it rude.
But I do not consider it a good thing that, after this superficial
comment
of yours, you did not even care to reply to my further arguments and
question.
This is precisely the kind of behavior that frustrates me and, I guess,
frustrates most other bug reporters.
This is not a superficial comment, my whole point with that comment
is that GCC does not __currently__ implement any other rounding mode
than the default one which is not what you want but hey it is what GCC
currently does.
You said something incorrect ("Note neg just flips a bit so it is
correct
anyways and there is no loss of precession"), then again you failed to
reply
to an explicit question and to further arguments I added. You also
failed
to reply when I pointed out, in another comment, that this was indeed
a regression from 3.3. Then leaving the bug UNCONFIRMED for months...
OK, you did not have time to check the standard... perhaps it is the
word "bugmaster" that generates unreasonable expectations.
That is because I said someone else needs to look into it (a C "lawyer"
person who is on the standards committee). Also again GCC
__currently__ implements one rounding mode and this is current for that
rounding mode. And actually neg does just a flip a bit according to the
standard. Now rounding modes are weird and should be implemented in GCC
but __currently__ they are not have not been since day one before there
was a C standard.
For the last one, the warning is correct, yes and is actually useful for
most people.
-- Pinski
