Hey @simulacrum, @centril, @Matthew Jasper --
when we last talked about migrating away from NLL, I think part of the plan was some kind of "hey! this is coming in 1.40!" sort of blog post. AFAIK this did not happen, right? Seems like something we should work on?
Or did we somehow convince ourselves that we've mitigated all of the concerning cases?
Channel Version Release Date
Stable 1.38 September 26th 2019
Beta 1.39 November 7th 2019
Nightly 1.40 December 19th 2019
Google says we have 23 days until Nov 7th
Yea would be good to write up that blog post; I am up for collaborating on it
Maybe also make a celebratory announcement that AST borrowck is byebye when 1.40 approaches?
Maybe @pnkfelix wants in
also cc @Pietro Albini
how's the current breakage situation?
@Pietro Albini we haven't checked after the last crater run
(which wasn't that long ago tho)
I think we should probably do a nightly crater run perhaps to try and check? Last I looked crater queue was ~5 days long
we have 5 PRs now
we lost one builder it seems
we are on distributed crater now
@simulacrum what does distributed crater mean exactly?
it goes much faster? :P
both agents/builders are working on the same job
cc @Pietro Albini
yep, we do half the experiments at the same time but a single one takes half the time
it doesn't really matter when there are queued experiments, but it will help in the future if there is only one experiment in the queue or we get more and smaller agents
wasn't really feasible before to get less powerful machines as the experiment on that agent would be pretty slow, but now we can get them as they'll just contribute whatever they can to the distributed experiment