On Mon, 12 Apr 2021 14:37:51 +0200 Jens Gustedt via Liaison wrote:
>
>Aaron,
>
>on Mon, 12 Apr 2021 07:59:03 -0400 you (Aaron Ballman
><aaron_at_[hidden]>) wrote:
>
>> Correcting the liaison email address so they're included.
>>
>> On Sun, Apr 11, 2021 at 5:12 AM Jens Gustedt <jens.gustedt_at_[hidden]>
>> wrote:
>
>> > Lambda expressions already have a interesting incompatibility with
>> > designators in initializers compared to C17, which I only found out
>> > by the integration mentioned above. If we are in a world where
>> > (implementation-defined) some constant objects can also be ICE, such
>> > as
>> >
>> > int const something = 42;
>>
>> <tangent>Perhaps I'm misunderstanding 6.6p10, but I don't believe that
>> gives you leave to turn this into an *integer constant expression*
>> just a *constant expression*. ICE is a more specific term that has
>> further semantic meaning elsewhere, and I was not under the impression
>> implementations were allowed to define new kinds of integer constant
>> expressions. e.g., I think it's invalid for an implementation to
>> decide these are identical declarations: int foo[something]; and int
>> foo[42];</tangent>
>
>interesting, I can read that differently.
>
>Such as that this phrase opens all the categories (ICE, address
>constants, ...) that are introduced in that clause. I vaguely remember
>that some implementation used that phrase to extend what they accept
>as address constant.

DR 312 against C99 has in the Committee Discussion:

Even if an implementation accepts other forms of constant expressions, paragraph 6.6#10 does not change the definition of an integer constant expression given by paragraph 6.6#6, and int[(int)+1.0] is still a VLA.

---
Fred J. Tydeman        Tydeman Consulting
tydeman_at_[hidden]      Testing, numerics, programming
+1 (702) 608-6093      Vice-chair of PL22.11 (ANSI "C")
Sample C99+FPCE tests: http://www.tybor.com
Savers sleep well, investors eat well, spenders work forever.