[PING^2] PR 23572 warnings for out of range floating-point constants

[PING^2] PR 23572 warnings for out of range floating-point constants

[Manuel López-Ibáñez]

The patch was rejected, as it is here
http://gcc.gnu.org/ml/gcc-patches/2007-01/msg01730.html
However, the conversation ended abruptly here
http://gcc.gnu.org/ml/gcc-patches/2007-01/msg01752.html
without specifying what the correct fix (if any) would be.
If no fix is possible, we should close PR 23572.

My proposal:

* Warning for overflow is pedwarn if infinities are not supported.

* If infinities are supported, warn for overflow with Woverflow.

* warn for underflow with Woverflow (so we don't have to introduce a
Wunderflow switch just for this).

I think this will be enough to close PR 23572.

Does this seem OK?

Cheers,

Manuel.

[PING^3] PR 23572 warnings for out of range floating-point constants

[Manuel López-Ibáñez]

Still pending...

On 10/02/07, Manuel López-Ibáñez <lopezibanez@gmail.com> wrote:
> The patch was rejected, as it is here
> http://gcc.gnu.org/ml/gcc-patches/2007-01/msg01730.html
> However, the conversation ended abruptly here
> http://gcc.gnu.org/ml/gcc-patches/2007-01/msg01752.html
> without specifying what the correct fix (if any) would be.
> If no fix is possible, we should close PR 23572.
>
> My proposal:
>
> * Warning for overflow is pedwarn if infinities are not supported.
>
> * If infinities are supported, warn for overflow with Woverflow.
>
> * warn for underflow with Woverflow (so we don't have to introduce a
> Wunderflow switch just for this).
>
> I think this will be enough to close PR 23572.
>
> Does this seem OK?
>
> Cheers,
>
> Manuel.
>

[PING^4] PR 23572 warnings for out of range floating-point constants

[Manuel López-Ibáñez]

Still pending resolution:
http://gcc.gnu.org/ml/gcc-patches/2007-02/msg00921.html

Thanks,

Manuel.

Re: [PING^4] PR 23572 warnings for out of range floating-point constants

[Joseph S. Myers]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 673 bytes --]

On Thu, 1 Mar 2007, Manuel López-Ibáñez wrote:

> Still pending resolution:
> http://gcc.gnu.org/ml/gcc-patches/2007-02/msg00921.html

I do not see what is supposed to be resolved by me.  I have already stated 
which cases are overflow or underflow (those arising from conversion of 
preprocessing tokens to tokens) and which are different and should be 
under -Wconversion (those arising from conversion of a value from one type 
to another), and which of the former should be pedwarns and which 
shouldn't.  I'll defer to a diagnostics maintainer on whether there should 
be separate -Woverflow and -Wunderflow.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: [PING^4] PR 23572 warnings for out of range floating-point constants

[Manuel López-Ibáñez]

On 01/03/07, Joseph S. Myers <joseph@codesourcery.com> wrote:
> On Thu, 1 Mar 2007, Manuel López-Ibáñez wrote:
>
> > Still pending resolution:
> > http://gcc.gnu.org/ml/gcc-patches/2007-02/msg00921.html
>
> I do not see what is supposed to be resolved by me.  I have already stated
> which cases are overflow or underflow (those arising from conversion of
> preprocessing tokens to tokens) and which are different and should be
> under -Wconversion (those arising from conversion of a value from one type
> to another), and which of the former should be pedwarns and which
> shouldn't.  I'll defer to a diagnostics maintainer on whether there should
> be separate -Woverflow and -Wunderflow.
>

Yes, you are right, the decision is whether there should be a separate
-Woverflow and -Wunderflow. Thanks for the clarification. Sorry for
the inconvenience.

Manuel.