Re: [netmod] Joint WGLC on "semver" and "module-versioning" drafts

[Kent Watsen <[e-mail]>]

Hi Martin!


> I think you meant https://github.com/netmod-wg/yang-next/issues/49.

Yes but, in spirit of the idea, I suppose both would be in play, if at all.


>> IMO the parsing of YANG files to produce a conceptual data model
>> is a critical component of the language itself.  Any statements that
>> affect this conversion step MUST be regular statements.
>> An extension is (by definition) an external statement that is not part of
>> the YANG language.
>> Critical extensions are not a good design choice.
> 
> I strongly agree.  Allowing the language to drastically change (like
> in this example) by just defining a critical extension is not a good
> idea.  Basically, anyone can define their own extension that changes
> YANG to whatever, and still claim conformance.

I also agree, for this issue (versioning), but there may be other places where a "critical" flag would help (wit the conversation that issue generated).  That said, for a limited-scope rfc7950-bis, it would be fine to not pick up either of these issues.


>> Just add real statements
>> instead.
>> 
>> IMO most of the yang-next issues are not interesting or valuable,
>> so a long WG process to go through this entire list is a non-starter.
>> 
> 
> The problem is that everyone will have their own pet-issues that they
> think are critical...
> 
> 
>> There are 3 critical changes that need to be made.
>>  - change "status" so deprecated and obsolete definitions are
>>    correct
> 
> ... for example I don't think this is critical ...
> 
>>  - introduce new instance-identifier data type based on RFC 7951 definition
>>  - introduce new identityref data type based on RFC 7951 definition
> 
> ... but I do agree with these!
> 
> 
> /martin


Whilst the chairs haven't closed this WGLC yet, I propose a YANG-next design team, asked to produce a limited-scope I-D they think best.  WG-objections of the form "my pet-issue isn't picked-up" should not be used to fail adoption (or, later, the WGLC).  Of course, objections to how the specific-issues picked-up were resolved are valid.  The goal being to most expediently (<1yr) forward the versioning work in a correct (contract-compliant) manner.  Support?

Kent