LWN's guide to 2024

Posted Jan 2, 2024 23:33 UTC (Tue) by iabervon (subscriber, #722)
Parent article: LWN's guide to 2024

My guess is that there will be a thread on Lore for a patch series where the developer didn't use a mail client to start it, read reviews, or respond to them, but other participants did receive this communication as email and send their feedback that way.

LWN's guide to 2024

Posted Jan 3, 2024 5:15 UTC (Wed) by mricon (subscriber, #59252) [Link] (4 responses)

I describe how I see a bridged GitHub/GitLab/Whatnot workflow for someone who wants to submit a pull request to a subsystem:

https://lore.kernel.org/tools/20231213-fluffy-roaring-cap...

It uses the email bridge as a mechanism to do additional code review and collect sign-offs from people who may not necessarily have an account on that particular forge.

I hope to work on this in the next few weeks.

LWN's guide to 2024

Posted Jan 3, 2024 11:19 UTC (Wed) by dullfire (guest, #111432) [Link] (3 responses)

I only took a quick skim, but it looks like this is primarily a github->mailing-list bridge (mostly unidirectional), is that correct?

Is there any tooling for if going the other way, that is I use a mail client (or basically any other interface that doesn't require a github account or agreeing to github's ToS) to contribute?

I am sort of guessing not (for one, I suspect that would itself be a violation of github's ToS), but I thought I'd ask.

LWN's guide to 2024

Posted Jan 3, 2024 20:49 UTC (Wed) by rincebrain (subscriber, #69638) [Link] (2 responses)

Github allows replying to PRs via email, so I don't think it _has_ to violate ToS...

LWN's guide to 2024

Posted Jan 3, 2024 21:22 UTC (Wed) by iabervon (subscriber, #722) [Link] (1 responses)

I think there's a federation issue, more than anything else: can kernel.org do anything that appropriately conveys that Al Viro reviewed the PR in a message kernel.org received from him? Or can kernel.org deliver Al Viro a useful email such that, if he replies in the usual way, the review arrives in the PR appropriately? (Al Viro being an example of someone whose normal email address is not at kernel.org.)

The forge and the email list probably have to be able to trust each other to have done the appropriate authentication of the attribution of review comments before transforming them in ways that preserve the meaning, which probably requires some sort of special agreement to be not too awkward.

I guess it might be okay to have the GitHub PR get a lot of review comments by kernel.org that say "according to (link to lore), (real person) says: (extracted comment about that part)", which kernel.org could add on its own behalf.

LWN's guide to 2024

Posted Jan 7, 2024 18:10 UTC (Sun) by mathstuf (subscriber, #69389) [Link]

I believe that every "issuable" has its own email address (using `+unique-id`-style addresses) and the sender is associated with the user account that sent the email (probably based on `From`, so mailing list header-rewriting might need to change). I don't know what happens if you send something to such an address without an associated GitHub account with the sending email address.