[BUGFIX release] upgrade rsvp + backburner

[stefanpenner]

[fixes #11469]

• add a bunch of tests (currently working on it)
• ensure we correctly don't throw if a catch handler was added asynchronously. I believe this worked at some point, but some refactorings have broken it. Now we just defer all the responsibility to RSVP, which is also now more aware of the run-loop (via configuration)

[stefanpenner]

@rwjblue r?

[rwjblue]

LGTM, likely going to stick with BUGFIX beta though (if you don't mind), that'll give us a little time to play with it before hitting stable.

[stefanpenner]

LGTM, likely going to stick with BUGFIX beta

the bug is in stable, but I'm fine with letting it sit in beta for some period then being backported to 1.13.x when its sufficiently vetted.

[rwjblue]

the bug is in stable

It is likely that all bugs that are unknown are in stable. 😺

I'm fine with letting it sit in beta for some period then being backported to 1.13.x

Everytime we backport something into a stable release we risk creating even more regressions, the bar for backporting to stable are (and should continue to be) very high.

We haven't updated Backburner or RSVP for quite a few versions (looks like RSVP hasn't changed since before 1.10 and the only changes in backburner that landed in 1.13 were for upgrading the build tooling). In other words, this isn't a new regression, it has existed in a number of stable versions (I am unsure if it has ever worked properly, actually), and is not a large show-stopping bug. I do not think it meets the requirements for backporting into the current stable branch.

[rwjblue]

Merging for pulling into beta ASAP.

@stefanpenner - Thank you for getting these updated and getting more of this stuff under test!

[jamesarosen]

@stefanpenner @rwjblue

the bug is in stable, but I'm fine with letting it sit in beta for some period then being backported to 1.13.x when its sufficiently vetted

If there's something I can do to help test this against 1.13, let me know. (I can try replicating the patch against 1.13.3, but I'd welcome any advice.)

[stefanpenner]

If there's something I can do to help test this against 1.13, let me know. (I can try replicating the patch against 1.13.3, but I'd welcome any advice.)

I suspect this will be a common trend during the transition period, maybe we need a 1.13.x-unstable or something branch.

I appreciate both sides 1. fix the bugs 2. don't introduce more bugs. Some balance will need to exist.

Luckily on the regular train we cycles to flesh out the most egregious issues early. But with 1.13.x and the future LTS releases we don't yet.

This actually may be something worth bringing up on the LTS rfc, it seems quite related.