Re: [Idr] Review of draft-abraitis-bgp-version-capability-07

[John Scudder <[e-mail]>]

On Aug 27, 2020, at 11:04 PM, John Scudder <jgs=[e-mail]<mailto:jgs=[e-mail]>> wrote:

one of the chairs will follow up tomorrow

(Co-chair hat on)

TL;DR: I’d prefer a different way forward, as a WG document, but I don’t see a basis for asking ISE not to publish if that’s the author’s preference. I do have a couple comments at the bottom that I’d like addressed.

Longer version:

I went back and reviewed the list traffic on this draft. Here’s the timeline I see:

- Donatas told the WG about the draft on August 2, 2019. This kicked off a thread that ran until August 20, 2019.
- In that thread, some people liked the draft, some didn’t, some suggested using draft-ietf-idr-operational-message to achieve similar functionality.
- The author never asked the list, or chairs, to adopt the draft as a WG item, so we didn’t do a formal call. In retrospect, we chairs should have been more proactive about this, since I see Donatas has only posted to the list one other time, which should be a hint that he might not be completely familiar with how we run the WG.
- Be that as it may, on November 21, three months after the previous thread trailed off, Adrian told the list Donatas had requested ISE publication. That was a second missed opportunity for the chairs to say “ahem”.

There was also some private traffic, in which Donatas sought guidance from the chairs and Adrian. I think maybe there was an in-person conversation between Adrian and me as well, that eventuated into Adrian’s November note to IDR.

If I had it to do over again, I would have stepped in after the August thread and encouraged Donatas to offer it as a WG document if that was his desire. I’d have provided feedback of that nature to Donatas’ unicast message. I’d also have stepped in after Adrian’s November message and said similar.

But I don’t have it to do over, and here we are. Donatas and Adrian now have sunk cost in publishing on the ISE stream. I also don’t know for certain what motivated Donatas to walk away after what, to me, was a fairly friendly and productive WG mail thread last August; I speak of this more a couple paragraphs down.

I don’t think it’s out of the question to adopt the work in IDR even at this late date; indeed there seems to be some interest in the problem space. My *guess* (and it is only a guess) is that the WG would end up preferring an approach that looks more like operational-message, though, and less like version-capability. I base this on the fact that the WG has been fairly steadfast in declining to bung strings into miscellaneous messages. I can think of several examples where it was proposed, and only one example where we’ve actually done it — RFC 8203, and I think the only reason that prevailed was that the session is already doomed in that case, so it’s easier to stomach doing something messy.

So, if Donatas is wedded to the exact implementation documented in draft-abraitis, then it is true, he shouldn’t offer it as a WG contribution, since that would hand change control to the WG. In that case, if the capability is to be shipped anyway, it’s better to document it with an ISE publication than to not document it. My impression from the mail I’ve reviewed is that this is the case.

On the other hand, if this was a misunderstanding and Donatas thought he’d been rejected and that’s why he took his document to Adrian… Donatas, you haven’t been rejected, you just didn’t ask in the way we were expecting to hear, and we chairs didn’t do a good job of encouraging your contribution. Sorry for that.

In short, I agree with Robert’s message of August 22, 2020. Procedurally, though, I don’t think IDR has a veto pen: we chose to give the registry an FCFS range, and by definition we don’t have a veto over that. The pointy end of permissionless innovation is, we don’t get to deny permission. I don’t think draft-abraitis would harm the BGP protocol, it’s just not the way the WG might choose to solve the problem. So, if Donatas wants to publish it as is (and it seems he does), then it’s settled.

I do have a couple further remarks about the draft, though:

First, I see that section 3 mentions that

   Although this document is not an IETF Standards Track document,

While I’m glad this is mentioned, I think it would be worth saying something similar in the abstract and introduction, instead of burying it in the middle. For example, “This document is not the product of an IETF working group and is not an IETF Standards Track document". Perhaps this is a usual thing to do with ISE stream documents anyway, and it would come later in the editing process, in which case sorry for the noise.

Second, a comment about naming: I find “version” to be a terribly confusing name, since BGP has a version number of its own. I would prefer almost any other name, for example “description” or the wordier “software-version”. “Version” on its own is ambiguous in a way I don’t think is helpful for anyone. (Feedback like this would have come as a matter of course, if the document had been a WG document.)

Thanks,

—John