Re: [nfsv4] Working Group last call for NFSv4.1 - ending September 23rd

[David Noveck <[e-mail]>]

> Please do note that most of these errata are the result of further
implementation experience by various vendors,
> and at this point, reflects code that has been in public use for several
years.

Good point. :-(

> As such, the default assumption should be that the errata are valid, and
any attempt to invalidate them needs
> strong justification,

I think the criterion that Magnus laid out, i.e. "changes a [wg] consensus
decision" is a sufficiently high bar for rejection.   Except for for one
small errata to which your point does not apply, any reccommendation to
reject.an errata would be made by this  working group and we should not
accept a bare assertion that a  consensus decision is changed, given that
this question is rarely cut-and-dried.  In my opinion, clear and convincing
evidence of this supposed change should be provided.

> since it would by implication also require changing the implementations
on which they are based

Note that if we do run into this sort of conflict, a bis document can
change a consensus decision reflected in the base document.  This is what
happened with internationalization in RFC7530.  So, even if we decide an
errata is to be rejected as changing a consensus decision, the editor of
the bis can indicate why, based on the existing implementations, the change
should be made.   However, any such decision will need to be justified to
the IESG, requiring a more detailed explamation than one that simply
reflects a verified errata..

On Tue, Sep 3, 2019 at 9:22 AM Trond Myklebust <[e-mail]>
wrote:

>
>
> On Sep 3, 2019, at 05:28, Magnus Westerlund <
> [e-mail]> wrote:
>
> My experience is that it is often quite easy to process the erratas.
> Someone with good knowledge of the RFC can usually determine if it is
> correct, wrong or needs WG discussion. So, yes I do expect that you can
> get through them within the timespan of a month.
>
>
> Please do note that most of these errata are the result of further
> implementation experience by various vendors, and at this point, reflects
> code that has been in public use for several years. As such, the default
> assumption should be that the errata are valid, and any attempt to
> invalidate them needs strong justification, since it would by implication
> also require changing the implementations on which they are based.
>
> Regards
>   Trond
>
>
> _________________________________
> Trond Myklebust
> Linux NFS client maintainer, Hammerspace
> [e-mail]
>
>