[cfp-interest 3386] Re: complex suffix specification

[RAJAN BHAKTA]

Looks good to me!

Regards,

Rajan Bhakta



From: cfp-interest <cfp-interest-bounces at oakapple.net> on behalf of Jim Thomas <jaswthomas at sbcglobal.net>
Date: Thursday, February 13, 2025 at 2:44 PM
To: CFP <cfp-interest at ucbtest.org>
Subject: [EXTERNAL] [cfp-interest 3385] Re: complex suffix specification
Based on discussion at the CFP meeting yesterday, here is a shot at content for a WG14 proposal … ----- 6. 4. 5. 3 #2 states “A floating suffix shall not designate a type that the implementation does not provide. ” The two not’s might be confusing. 

Based on discussion at the CFP meeting yesterday, here is a shot at content for a WG14 proposal …

-----

6.4.5.3 #2 states “A floating suffix shall not designate a type that the implementation does not provide.” The two not’s might be confusing. If the implementation supports decimal types, does 1.0dfi present a constraint violation? The combination dfi does not describe a type because C does not specify complex decimal types. One might think the constraint that the suffix “shall not designate a type that the implementation does not provide” is satisfied because the combination of suffixes doesn’t designate any type. A suggested change below is intended to express the intent unambiguously. (If the implementation does not support decimal types, the df suffix presents a constraint violation, as noted in footnote 66.)

In 6.4.5.3 #5, the first two sentences are about complex suffixes. The rest of the paragraph is about something else, namely unsuffixed literals and literals with real floating suffixes, and should be in a new paragraph.

Suggested changes:

In 6.4.5.3 #2, replace

A floating suffix shall not designate a type that the implementation does not provide.

with

A floating suffix shall designate a type that the implementation provides.

In 6.4.5.3 #5, introduce a new paragraph after the first two sentences, i.e. beginning with

An unsuffixed floating literal ...

-----

On Feb 6, 2025, at 9:49 AM, Jim Thomas <jaswthomas at sbcglobal.net> wrote:

The following comments are about specification of complex suffixes in N3435.

6.4.5.3 #5 The third sentence, beginning with "An unsuffixed floating literal …”, should begin a new paragraph.

6.4.5.3 #2 Is 1.2dfi a constraint violation? Footnote 66 says yes if the implementation does not support decimal types. But this should be implied by the normative specification. What if the implementation supports decimal types? The suffix dfi does not describe a type (because C does not specify complex decimal types), so it satisfies the constraint that it "not designate a type that the implementation does not provide.” Consider changing to “designates a type that the implementation provides.”

- Jim Thomas



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20250213/c78c692a/attachment-0001.htm>