Re: A query about the conformance questionnaire
On 2015-03-04 08:46:54 +0000, John Pryce wrote:
> I raised an I/O point with Christian, below. In the text on Input should we:
>
> - Say the rule for "snipping" a stream is language-defined (which I guess
> is the usual case, rather than being implementation-defined?) and shall
> be documented by the implementation?
> - Say the rule is implementation-defined, and shall be documented?
> - Specify such a rule, e.g. by saying it shall conform to what some
> popular language does by default?
> - Say nothing about this rule, as at present?
P1788 only defines conversion from/to *strings*, not streams. Any
extension to streams is out of scope of P1788 and is typically
language- or implementation-defined, which is basically what 13.1
says.
> He is right that if the main text says nothing about it, it can't
> appear in the conformance clause.
I agree about streams. Concerning the locales, this is an important
point, because in particular, case matching (referenced in P1788)
depends on the locales, and the behavior in languages is not always
clear. For instance, consider a program that uses the literal "[NaI]"
and the program is *compiled/interpreted* under Turkish locales (for
which this dotless "I" is not the uppercase "i"). Or consider "[NaI]"
provided as input when a program in *run* under Turkish locales. In
both cases, the behavior must be clearly specified by the language
and/or the implementation.
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)