Ah, I see where I misread the paper. The summary section indicates that "Clang" would be OK for 32-bit values. Clang exists on Windows with 16 bit values and on Linux/Mac with 32-bit values. Perhaps indicate the platform instead of the compiler?

Thanks for taking the suggestions into consideration,
Peter

On Wed, 14 Apr 2021 at 17:05, Peter Brett via SG16 <sg16@lists.isocpp.org> wrote:

Hi Peter,

 

Just to follow up, a key takeaway from your feedback is that the table in the “Summary” section is misleading. Thanks for highlighting it. I’ll improve it before the mailing goes out.

 

Thanks again,

 

                  Peter

 

From: SG16 <sg16-bounces@lists.isocpp.org> On Behalf Of Peter Brett via SG16
Sent: 14 April 2021 15:47
To: Peter Bindels <peterbindels@gmail.com>
Cc: Peter Brett <pbrett@cadence.com>; sg16@lists.isocpp.org
Subject: Re: [SG16] P2362R0 Make obfuscating wide character literals ill-formed

 

EXTERNAL MAIL

Hi Peter,

 

Thanks for your feedback. I’m looking forward to your counter-proposal that makes the Windows situation well-formed.

 

Best regards,

 

                        Peter

 

From: SG16 <sg16-bounces@lists.isocpp.org> On Behalf Of Peter Bindels via SG16
Sent: 14 April 2021 15:24
To: Corentin Jabot <corentinjabot@gmail.com>
Cc: Peter Bindels <peterbindels@gmail.com>; SG16 <sg16@lists.isocpp.org>
Subject: Re: [SG16] P2362R0 Make obfuscating wide character literals ill-formed

 

EXTERNAL MAIL

I value portable code highly. If we have known platforms, and a competing proposal that makes the Windows situation well-formed, it makes this an odd sell to me. I do not want a specification that requires code to compile on Unix and fail on Windows, even if it is not currently according to spec and somehow unportable code.

 

Blocking multi-character literals, sure. Blocking some single-character literals as IFDR when they work fine on other platforms, that's just not right. I don't think I like either variant - making it ill-formed on Unix, or to allow it on Unix while making it ill-formed on Windows.

 

On Wed, 14 Apr 2021 at 16:16, Corentin Jabot <corentinjabot@gmail.com> wrote:

 

 

On Wed, Apr 14, 2021 at 4:07 PM Peter Bindels <peterbindels@gmail.com> wrote:

Then I have the inverse problem, now we create code that is conditionally portable, and only ill-formed on Windows. That's horrible.

 

How is that worse that code that is well-formed everywhere but does the wrong thing on windows?

wchar_t was never portable

 

 

On Wed, 14 Apr 2021 at 15:55, Corentin Jabot <corentinjabot@gmail.com> wrote:

 

 

On Wed, Apr 14, 2021 at 3:49 PM Peter Bindels via SG16 <sg16@lists.isocpp.org> wrote:

Please explain how that facepalm is now illegal on Unix, where wchar_t is 32 bit and it clearly fits.

 

It's only ill-formed if it doesn't fit!

 

 

On Wed, 14 Apr 2021 at 15:27, Peter Brett via SG16 <sg16@lists.isocpp.org> wrote:

Hi all,

Corentin and I have authored https://isocpp.org/files/papers/P2362R0.pdf, which will be in the April mailing.  This addresses https://github.com/sg16-unicode/sg16/issues/65.

Best regards,

                Peter
--
SG16 mailing list
SG16@lists.isocpp.org
https://lists.isocpp.org/mailman/listinfo.cgi/sg16

--
SG16 mailing list
SG16@lists.isocpp.org
https://lists.isocpp.org/mailman/listinfo.cgi/sg16

--
SG16 mailing list
SG16@lists.isocpp.org
https://lists.isocpp.org/mailman/listinfo.cgi/sg16