On 26/06/2018 16:41, Juergen Schoenwaelder wrote:
On Tue, Jun 26, 2018 at 05:38:10PM +0200, Balazs Lengyel wrote:
Any opinions on Rob's suggestion about a free-text versioning string?
I am neutral on this.
It needs to be clear _what_ is versioned. I have seen conflicting
views. I have heard so far:
- the YANG version context once (module, path) is not sufficient anymore
I agree with Juergen, that depending on where YANG versioning goes, that
a mechanism to provide this information is required (e.g. a list of
modules + semvers for the schema used in the instance data). It should
be optional to publishers whether they populate this information.
- the instance data version
The draft already has instance-data revisions. I regard that as a
mechanism for versioning instance data. I'm not saying that this is
necessary the right approach.
- a version number refering to a specific software release of a server
I see this "version" as file level meta-data information, in the same
category as name, contact, description, organization meta-data fields
that the draft currently provides. Another field that may be
generically useful is the time/date stamp of when the instance data was
generated. Although perhaps the timestamp of the file is sufficient for
this purpose.
Thanks,
Rob
/js
_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod
