PATCH PING: Make canonical type failures ICE

PATCH PING: Make canonical type failures ICE

[Doug Gregor]

This simple patch turned canonical type failures into ICEs, so we can
weed out the remaining problems with the canonical type system:

  http://gcc.gnu.org/ml/gcc-patches/2007-07/msg00331.html

Okay for mainline?

  - Doug

Re: PATCH PING: Make canonical type failures ICE

[Gabriel Dos Reis]

"Doug Gregor" <doug.gregor@gmail.com> writes:

| This simple patch turned canonical type failures into ICEs, so we can
| weed out the remaining problems with the canonical type system:
| 
|   http://gcc.gnu.org/ml/gcc-patches/2007-07/msg00331.html
| 
| Okay for mainline?

Unlike the description you gave above, the patch really is not
strictly about turning the warning into an ICE -- a long standanding
request with multiple bugzilla entries.

The part of the patch that turns the warning into an ICE is approved.
You might want either:
  (1) to split the patch so that each addresses on point
  (2) or to have someone look at the non-"ICE" part of your patch.

-- Gaby

Re: PATCH PING: Make canonical type failures ICE

[Doug Gregor]

On 26 Jul 2007 15:55:44 -0500, Gabriel Dos Reis <gdr@cs.tamu.edu> wrote:
> "Doug Gregor" <doug.gregor@gmail.com> writes:
>
> | This simple patch turned canonical type failures into ICEs, so we can
> | weed out the remaining problems with the canonical type system:
> |
> |   http://gcc.gnu.org/ml/gcc-patches/2007-07/msg00331.html
> |
> | Okay for mainline?
>
> Unlike the description you gave above, the patch really is not
> strictly about turning the warning into an ICE -- a long standanding
> request with multiple bugzilla entries.

It renames a --param that affects when checking of canonical types is
done, because the previous parameter name and behavior makes no sense
with the change to an ICE. All of the above was explained in the
message I linked to.

>   (2) or to have someone look at the non-"ICE" part of your patch.

There's no point in splitting the patch, so I'll wait on this.

  - Doug

Re: PATCH PING: Make canonical type failures ICE

[Gabriel Dos Reis]

On Thu, 26 Jul 2007, Doug Gregor wrote:

| On 26 Jul 2007 15:55:44 -0500, Gabriel Dos Reis <gdr@cs.tamu.edu> wrote:
| > "Doug Gregor" <doug.gregor@gmail.com> writes:
| >
| > | This simple patch turned canonical type failures into ICEs, so we can
| > | weed out the remaining problems with the canonical type system:
| > |
| > |   http://gcc.gnu.org/ml/gcc-patches/2007-07/msg00331.html
| > |
| > | Okay for mainline?
| >
| > Unlike the description you gave above, the patch really is not
| > strictly about turning the warning into an ICE -- a long standanding
| > request with multiple bugzilla entries.
| 
| It renames a --param that affects when checking of canonical types is
| done, because the previous parameter name and behavior makes no sense
| with the change to an ICE.

Obviously, the change to ICE can be made without changing the fiddling
with the interface to the compiler, as can be read from the message
you linked to.

-- Gaby

Re: PATCH PING: Make canonical type failures ICE

[Jason Merrill]

Doug Gregor wrote:
> This simple patch turned canonical type failures into ICEs, so we can
> weed out the remaining problems with the canonical type system:
> 
>  http://gcc.gnu.org/ml/gcc-patches/2007-07/msg00331.html
> 
> Okay for mainline?

Mark approved this patch on July 10.

Jason