Re: An IETF repository for working code in our protocols?

[Vijay Gurbani <[e-mail]>]

Dear Mark: Nice to hear from you, and thanks for your comments.  A couple
of observations inline.

On Wed, Aug 19, 2020 at 8:30 PM Mark Nottingham <[e-mail]> wrote:

> Hi Vijay,
>
> Just my .02 - I'd be wary of an IETF repository of such code, as that
> would convey too much authority on it. It would also create a fair amount
> of overhead (especially for long-term maintenance).
>

Conveying too much authority is a valid concern, no doubt.  As others have
expressed, we do not want to condone a "reference implementation", or
engender some type of advertisements by submitting source code.   This
would have to be worked out through a process, if we get that far.
Regarding overhead for long-term maintenance, the effort of maintenance
could be made similar to the effort we spend on maintaining RFCs, where
most of the effort is in filing and validating errata.  Errata for source
code would mean a patch, and I can see supporting a patch or two for the
code that was submitted when the archive was established by the publication
of the RFC.  However, I do not foresee that we would constantly keep adding
new implementations to the archive, that should be out of bound.


> I also don't think that it's necessary, once a protocol is successful;
> after broad deployment, people don't need to go searching for
> implementations. It *is* useful when a project is starting, to help early
> implementers find each other, though.
>

Yes, agreed!  The idea would be light the fire of widespread implementation
by the kindling (code) that tracked the I-D as it became an RFC.


> In the WGs I participate in, we often use wiki pages to track early
> implementations; e.g.:
>
> * https://github.com/httpwg/wiki/wiki/Structured-Headers
> * https://github.com/quicwg/base-drafts/wiki/Implementations


The problems with Wiki pages and datatracker hosted additions is that we
(i.e., the IETF community) know what these are and where to find them.  But
I maintain that the average developer who is implementing an RFC will not,
on the average, have the required background to know where in the ietf.org
domain it should go to in order to find all of these additions.  They will
only know one thing, the RFC.  They will know the errata associated with
the RFC because a link to the errata is on the RFC title masthead.  It will
be great if there was a link to some starter code for new protocols.

(I suspect that I have not convinced you, but that is okay.  This is an
exchange of ideas so far to see how much of this is indeed workable.)

Thanks!

- vijay