Re: [PATCH 2/7] Docs: Bring SubmittingPatches more into the git era

From: David Woodhouse
Date: Wed Mar 09 2016 - 10:16:24 EST



> I wrote that text that way because certain high-profile maintainers have
> said exactly that sort of thing:
>
> You can send me patches, but for me to pull a git patch from you,
> I need to know that you know what you're doing, and I need to be
> able to trust things *without* then having to go and check every
> individual change by hand.
>
> -- Mr. T. https://lwn.net/Articles/224135/
>
> ...and because, in truth, few maintainers do take pull requests. There
> *is* some value in having the code out on the lists in the clear, it
> raises the chances of somebody *else* looking it over slightly. There is
> a reason why review is done on the lists, not directly from repositories.
>
> Allowing the maintainer to attach tags certainly seems like another valid
> reason to defer setting patches into git-implemented stone. But I don't
> see it as the only one.
>
> We could, I suppose, run a poll to ask maintainers why they are reluctant
> to take pull requests. But the end result is kind of the same as far as
> readers of SubmittingPatches are concerned - they need to send their
> patches via email.

You are quite right that it has the same effect in practice, for Linux.
The problem was that your words were being taken out of context in a
situation where email review *was* always going to be required anyway, but
I'm trying to get them to allow pull requests instead of always losing
history by *forcing* a rebase onto the current HEAD.

Which is a model we use often too -- post for review and feedback, but
submit a pull request with the *actual* set of commits that were tested,
on the base they were developed against. Instead of submitting *only*
patches and running the risk that what gets committed to today's tree has
*never* actually worked correctly, when we look back at the inaccurate
history.


--
dwmw2