PHP floating point bug possibly misinterpreted

[Andrew Haley]

On 01/14/2011 02:29 PM, Michael Cronenworth wrote:
> Andrew Haley wrote:
>> It's perhaps important to note that gcc is not standards-conforming
>> by default, and if you want it to be you have to use the appropriate
>> options to say which standard you want it to conform to.  The
>> defaults are a best guess at what people might want.
>
> That's all fine and dandy, but I obviously did not make myself clear due
> to the responses I have received.
>
> Fedora turns on some C flags *by default* for security purposes, and I
> felt that this flag might need to be added to that list to prevent
> future (unintended) security holes.

OK, I see.

> Is there a real use case for ever requiring the undefined nature
> of the default behavior?

Absolutely, yes.  This is all discussed at fantastic length in the
references mentioned.

In summary: All workarounds for this bug are problematic.  Jim Wilson
nicely summarizes here: http://gcc.gnu.org/ml/gcc/2003-08/msg01195.html

One group or posters considers any failure in this area to be
extremely serious, so much so that it does not matter how much
efficiency is lost, the bug must be fixed.  The other group says that
this has all been understood for many years, and in the few cases
where it really matters workarounds can be applied.  Each group is
sure that the other group is crazy.

As rth put it,

> > But wouldn't it be better than current situation?
>
> No, not better, just different.  You anger a different set of folk.

Andrew.