BIP3: clarify number assignment

[jonatack]

There's been confusion about this, so add the following clarification:

"A number may be considered assigned only after it has been publicly announced in the pull request by a BIP Editor."

The following, for instance, should not constitute assignment of a BIP number:

• an announcement on social media
• a provisional entry to the internal editor notes, pending feedback from the other editors (the entry can be subsequently removed)

As well, BIP 2 already stipulates that assignment take place in the pull request:

* Assign a BIP number in the pull request.

[ajtowns]

Concept +1, but, alternatively, you could publish a canonical list of assigned numbers eg as a separate git repo (in which case editors could propose assignments as PRs) or as a wiki page (in which case proposed assignments would need to be out of band). That might make it slightly easier to go from "oh BIP xxx is assigned but not merged, which PR# was that again?"

[jonatack]

We do have a private git repo to coordinate assignments. Initially we opened PRs there, but it turned out that colleague feedback was easier to obtain in a different way. It was also suggested fairly early on to push to the repo directly. This worked fine until recently, but given recent events, I think it is best to clarify in BIP 3 at what moment a number is assigned, and IMO that moment should be publicly on the PR, like BIP 2, as proposed here and as I think we generally see it.

[luke-jr]

Weak concept NACK. This prevents issuing numbers to authors directly, which has historically been done. OTOH, it hasn't been the practice recently since the de facto requirement for a PR, so maybe it's fine. But in that case, Rationale should probably mention/justify it.

[jonatack]

Note that BIP 2 already stipulates that assignment take place in the pull request.

* Assign a BIP number in the pull request.

[murchandamus]

LGTM. This matches both what I considered to be the current modus operandi, and what I intended to happen for BIP 3. Thanks for catching that I overlooked stating it explicitly.

[murchandamus]

Weak concept NACK. This prevents issuing numbers to authors directly, which has historically been done. OTOH, it hasn't been the practice recently since the de facto requirement for a PR, so maybe it's fine. But in that case, Rationale should probably mention/justify it.

This practice appears to have led to several number assignments to ideas that never substantiated into written documents. It’s not clear to me why it would ever be necessary or advantageous to assign a number privately.
Both BIP 2 and BIP 3 describe that a number assignment follows a properly formatted, high quality draft being submitted per a pull request, so it’s not undue to expect that a pull request exists at the time of a number assignment, and the corresponding pull request is obviously the appropriate place to announce the assignment (even explicitly stated in BIP 2).