[RFC] Should we merge these backports?

Putting on my core contributor hat. The following backport PRs have fallen on my plate, and as per the build-test-release group’s latest ADR, I’m bringing them up for comments, defense, and/or objections.

The discussion can happen here or in the individual issues, as you see fit. If no objections are raised until our next meetup, I’ll raise the point during it, and if that does not produce objections, I’ll consider merging the PRs as core contributor.

I’ll kick off the discussion with a question: should we ever merge anything back to an unsupported release? We seem to have converged on keeping N-1 (and whatever is running on CI) installable, but backporting a feature to Koa might be a step too far.

Particularly interested in @BbrSofiane’s, @cmltaWt0’s, @regis’s, and @nedbat’s opinions, as well as the original author’s, @gabor, @0x29a, and @Agrendalath.

I see two pull requests that introduce new, complex features and alter the old Python API. I do not understand why any of these PRs should be merged in the current release branch, let alone in the former one.

1 Like

Fair enough, @regis. I agree, but needed somebody else to judge.

To be clear, though, does the objection apply to the last PR as well? That seems to be lower risk than the others.

And while we’re on the subject, another backport PR sprang up:

This is pretty simple, but it does alter default logging behavior. Does it pass?

Tagging @mahyard as the original author.

Thank you @arbrandes,
I’m not sure if I am supposed to give an answer here. though, as a defense, I want to ask you to consider that my PR is mostly like a bugfix. As it removes some unworthy lines that were being logged very often (each time a learner requested a video without a transcribe). we tried to keep all useful information and I believe that not only it’s not going to interrupt any routine of the community members, but also it would improve their system’s performance by significantly reduce their I/O.