On 5/5/22 11:05, Carsten Bormann wrote:
>>>> * Would it make sense to further refine your contact leafs to check for 
>>>> the MUST URI schemas?
> (They are URI schemes, not URI schemas.)
>
> It is generally a mistake to hardwire specific URI schemes in a specification 
> that uses URIs to provide flexibility.
> (What if in a few years you would want to use a newly popular instant 
> messaging URI scheme for contacting people?)

This is a fair point.  "tel" may be "holo" in the future...

>
> But let me go with the program.
>
>>> I know what you want, but I'm not quite sure it is well specified how to do 
>>> it.  Patterns in general seem underspecified.  If someone can explain how 
>>> to do it, I'm happy to add.
> I’m not sure whether you are saying YANG patterns are underspecified:  They 
> are not.
>
> https://www.rfc-editor.org/rfc/rfc7950.html#section-9.4.5
>
> If you want to know more about the type of regexp that is being used there, 
> you may want to check 
>
> https://www.ietf.org/archive/id/draft-ietf-jsonpath-iregexp-00.html#name-xsd-regexps
>
> (Section 5 tells you how to map the iregexp subset into regexp dialects you 
> may be more familiar with.
> While iregexps restrict the regexps a little further than YANG does, the 
> iregexp subset is usually what you actually need and also is essentially the 
> subset that actually works the way many people would think they do.
> If you see a non-iregexp YANG pattern, I’ll bet you a beer that this is 
> actually broken [1].)
>
>> I can't recall having seen this done before,
> (Maybe because premature restriction of the schemes is, after all, not such a 
> bright thing to do?)

Indeed, but the normative language is there already, the typedef says
it's allowed, and I thought it might be better to also codify this.

>
>> but I was thinking along the lines of:
>>
>> type inet:uri {
>>   pattern "(mailto)|(https?)|(tel):.*";
>> }
> Wrong precedence; I think you want
>
>>   pattern "(mailto|https?|tel):.*";


Yes, sorry.

Joe
_______________________________________________
OPSAWG mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsawg

Reply via email to