Re: Running code (Was: I-D Action: draft-ietf-6man-ipv6only-flag-03.txt)

[Job Snijders <[e-mail]>]

On Tue, Oct 23, 2018 at 10:03:35PM +0200, JORDI PALET MARTINEZ wrote:
> Let’s suppose there is a draft and nobody wants to implement it.

full stop. then the document does not need to be published.

> Another draft is implemented. May be after some time the one that was
> not implemented get more market success than the other one.

fantastic, at that point the document can move forward again in the
publication process. 'market success' implies running code came into
existence.

> This is discriminating both drafts in the IETF process, and we must
> not do that. It may happen even “you have programmers” that are good
> friends, they will do for me and in the other way around. Or I’m a
> vendor, so I’ve the resource to do it.

This is not discrimination. If authors don't have the capability to
develop running code themselves, and also don't have access to resources
nor are able to convince others to implement a protocol specification...
the IETF's prime directive of interoperability can't be met anyway.

Some ideas gain traction, some don't - this happens to all of us.

> This will mean at the end that we balance the IETF work towards those
> that can (somehow), get things implemented, even if is a bad work (I
> understand that still need to follow the consensus, etc.).

Bias towards getting things implemented is an excellent way to foster a
stream of healthy specifications. Invariably the creation of running
code leads to higher quality specifications as implementers provide
their feedback to the WG. We don't work in the void, we're here to do
tangible things. This is not an academic exercise.

> At the end, we may have open source implementation of a draft, and not
> being adopted by vendors, or in the other way around. That’s not a
> good way to classify “good” or “bad” IETF work, also because after
> some years, what was “bad” is adopted as good, and viceversa.

I don't understand your 'open source' vs 'vendors' statement, they are
the same, both are vendors. The publication license and cost are
irrelevant.

> I understand that when IETF started, it was very few work and it was
> very easy to ask for code for everything, but our industry changed and
> we can’t enforce that, even at WG level.

I don't think 'our industry changed' means that we can now do without
code. We can absolutely enforce it, also at the WG level. IDR working
group is living proof. The QUIC and TLS13 groups have done excellent
work and only got to the point where they are by investing time &
resources to produce lots of running code.

Raising the bar is a good thing!

Kind regards,

Job