Re: [netmod] Balazs Review of draft-ma-netmod-with-system-02

["maqiufang (A)" <[e-mail]>]

Hi, Balazs,

Please see my reply inline.
From: Balázs Lengyel [mailto:[e-mail]]
Sent: Friday, April 1, 2022 1:22 AM
To: maqiufang (A) <[e-mail]>; Jan Lindblad <[e-mail]>
Cc: '[e-mail]' <[e-mail]>
Subject: RE: [netmod] Balazs Review of draft-ma-netmod-with-system-02

Hello Qiufang,
While it is possible to understand the draft in a correct way, I still believe that a few things should be much more explicit, would require strong straight statements:
- Definition of what is system configuration
- Update of <system> datastore has no effect on the <running> datastore even if some of it <system> has been copied over to <running> earlier.
- Contents of <running> are modifiable or not following the general YANG and NACM rules even if some of it <system> has been copied over to <running> earlier. Any further restriction will be defined in a separate “immutable” draft.
Etc.
Sure, the authors will update the draft to reflect the latest discussion and thoughts.

I understand that the immutable draft can work without the system draft, but I  don’t know if the system draft can work without the immutable draft. What is your opinion?

Yes, the immutable draft can work without the system draft.
Without immutable draft, the client may encounter some errors if it attempts to update a non-modifiable system-defined node. But the server today could reject any unreasonable request (more than just attempts to modify read-only data nodes) in the real world.
While immutable draft tries to make immutability information transparent, it’s hard to say that it can avoid all potential error responses but only help identify and understand why *some* of the edit requests are denied.
So I think it is okay for system draft to work without the immutable draft. But immutable draft does help complete the with-system solution.

Regards Balazs

Best Regards,
Qiufang