Re: [Netmod-ver-dt] Balazs Review of draft-verdt-netmod-yang-semver-01

["Joe Clarke (jclarke)" <[e-mail]>]

On Sep 13, 2019, at 09:34, Balázs Lengyel <[e-mail]<mailto:[e-mail]>> wrote:

Hello Joe,
Some comments:
General: I would like this to be merged with module-versioning.

  *   I would like this to be main-stream. It may be optional, but it should be the single recommended scheme.

I like keeping this separate and having the label schemes modular.  This is especially true when you consider other consumers like packages.


  *
  *   Putting restrictions on “other” schemes that use the revision label can not be enforced. If they don’t support this draft/RFC they will never know what they are not allowed to do.

That’s fair.  This text is a restatement of what will also be in module-versioning, and may become just a reference to there.  If one uses the revision label concept, then, they would know that the semver pattern is reserved.


  *
  *   A YANG module with a single typedef looks really strange.

I agree, but maybe no more odd than a single extension statement.  We discussed perhaps making this part of the module version module, but since you have packages as consumers, breaking it out might be okay.  We saw similar things when they packed yang-data into RESTCONF.

I’ll review your inline changes as I make further changes to the text today.  Thanks for the review!

Joe


  *

After workgroup adoption I would raise this issue again.

Ch 1)
Add something like:
“The goal is to add a human readable version label that provides compatibility information for the module without comparing or parsing the module’s body.”
This is the whole reason to create this draft.

Ch 2. And General) We really need to give our versioning scheme a “name” and refer to it as such. I propose “yangver”.

Ch 3.1 paragraph after the bullets)
Do we need to specify different metadata  types? Or is this text directly from semver?

Reword as it is hard to understand:
Old text:
   Versioning schemes that are not this modified semver MUST NOT use
   this same version pattern.  Other schemes may, for example, use
   additional leading characters to distinguish them from this semantic
   versioning scheme.
New text:
   Other version schemes MUST use versioning strings that do not match the format and the pattern specified for yangver , e.g. use
   additional leading characters to distinguish them from yangver.

Ch 3.4.2) I have not read the package draft recently, but does this example imply that a package has a single revision-label? IMHO for each revision a description, reference and revision label is needed.

Ch 4) Shouldn’t we allow any BC version to be used for import even if it is not in the revision history.

Ch 5.) I would find a more descriptive name for the typedef e.g. yangver

Regards Balazs

--
Balazs Lengyel                    Senior Specialist                       Ericsson Hungary Ltd.
Mobile: [phone]              email: [e-mail]<mailto:[e-mail]>

_______________________________________________
Netmod-ver-dt mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/netmod-ver-dt