On Sun, Jul 20, 2025, at 3:54 AM, Niels Dossche wrote: >> When is a discussion allowed to be considered wound down?
> This has never been defined and always has a subjective part. > In general, no one will block you from bringing it to a vote if no > substantial changes have been made to the RFC and the discussion lasted > at least 2 weeks. Indeed, and I've been yelled at in the past for making non-trivial changes to an RFC "too close" to when the vote is called. (For some undefined definition of "too close.") > Note though that the fact that the RFC still includes this does show a > non-consensus from the authors PoV. > Either you fully stand behind your own RFC and wouldn't have split the > vote, or you agreed that the get hook is a bad idea. > In the latter case, why even include this still in the RFC text, > especially after Larry said he's positive that part won't pass? > This comes across as really wanting something of the RFC to pass, not > aiming for the best "solution". No, we split the vote because, as stated, based on the available evidence (this list) "set" appears to be uncontroversial, but "get" is. But ripping out half the RFC right before calling a vote would certainly qualify as a not-small change, and if we want to get the "set" portion into 8.5, we have a short window before a vote can be called. Hence, splitting the vote, which is the smaller change. Submitting essentially a new RFC at this point isn't really an option. If by some miracle the 'get' hook also passes, I'm OK with that. But it's hardly the first vote that's been started despite considerable opposition. I do not recall seeing anyone make a compelling argument against readonly 'set' hooks, so at the moment it does look to us like allowing set hooks, at least, is a "best solution." (Or at least, best presented so far.) The idea that a split vote means the authors don't support their own RFC is nonsensical, given that multi-part votes or split votes or secondary votes come up multiple times every release. Rather, it means the different parts are related enough and small enough to be digestible in a single RFC, but each can stand on their own if necessary. In this case, nothing about the get hook necessitates a set hook, and nothing about the set hook necessitates the get hook. --Larry Garfield
