Re: [netmod] NBC changes and warnings

[Andy Bierman <[e-mail]>]

On Wed, Apr 20, 2022 at 7:51 PM Fengchong (frank) <
[e-mail]> wrote:

> Andy,
>
>        Server don’t know how the client’s yang modules is, so server can
> not judge whether the NBC changes exist.
>
>
>

The client does not have YANG modules.
The client loads the modules to match the server's YANG library.

YANG status-stmt is not correct, but the current best practice in the
software world
for NBC changes.  (1) Warn First (2) Then Change.
A deprecation warning (quite common in modern tools) means
"This operation still works now, but it will change or go away soon".

Andy


*发件人:* netmod [mailto:[e-mail]] *代表 *Andy Bierman
> *发送时间:* 2022年4月21日 0:30
> *收件人:* NetMod WG <[e-mail]>
> *主题:* [netmod] NBC changes and warnings
>
>
>
> Hi,
>
>
>
> Before any significant NBC change can be introduced, a process needs
>
> to be in place to minimize the disruption that the NBC change will cause.
>
>
>
> Consider the recent change by GitHub to remove support for SSH passwords
>
> and require keys instead.  They warned people for about 2 years this change
>
> was coming.  Not just on the WEB site, but in the tools. Every time
>
> a user logged in over SSH with a password, a warning was printed.
>
>
>
> For those willing to read the warning and act upon it, there was no
>
> service disruption at all when SSH passwords were finally turned off.
>
>
>
> When NETCONF was introduced in 2006, everything was new.
>
> The NETCONF error-severity=warning did not even work, and nobody cared.
>
> It still doesn't work.
>
>
>
> There are other possibilities besides
>
> "We can never make an NBC change because nobody will read the warning
>
> in the description-stmt anyway."  That's true. So let's fix it in the
> protocols
>
> and in the YANG language, so they do get the warning.
>
>
>
>
>
> Andy
>
>
>