Re: [core] [art] [Last-Call] Call for comments on draft-ietf-core-problem-details-07

[Carsten Bormann <[e-mail]>]

> 
> Yes, YANG needs to do what is right for YANG and not put back pressure on other specifications.  If and when there is a proposal for a YANG type for language tags, I will push for the lower case ie as Carsten says
> 
> "So I would have used [a-z]{1,8}(-[a-z0-9]{1,8})* instead of [a-zA-Z]{1,8}(-[a-zA-Z0-9]{1,8})* as the regexp for a language tag."

Yes, that is a complete no-brainer UNLESS there is stuff out there that prefers something different.
Actually, the capitalization rules on 5646 aren’t that complex, so they could be part of a presentation engine for language tags that are interchanged in lower case form.

> but I see that as one for the NETMOD WG perhaps with input from art if some want mixed case:-( YANG can be a but flaky when it comes to string comparisons.  

Note that we are talking about ASCII here ([a-z], that is).
The concept of “case-insensitive” is essentially unworkable outside this assisted living space.

> Some places it insists on lower case (e.g. 'true' 'false'), others it does not. 

But that is not very relevant to *using* strings in YANG.

>  It does have canonical forms but does not do Unicode normalisation and does allow a very wide range of characters.

If you do \d or \w in your patterns, you are!
(Which is one reason why they aren’t in iregexp [1], which is a proper subset of YANG patterns and CDDL regexps that actually is likely to work in reality.)

Don’t do that, then [2].

[a-z] and [0-9] are unambiguous.

Grüße, Carsten

[1]: https://www.ietf.org/archive/id/draft-ietf-jsonpath-iregexp-00.html
[2]: https://memegenerator.net/img/instances/80970645/doctor-it-hurts-when-i-do-this-then-dont.jpg