[plug] Kernel development process

[Daniel Hartwig]

On 12 July 2012 13:42, Andrew Cooks <acooks at gmail.com> wrote:
> Hi List
>
> Can someone explain to me why the kernel patch submission
> process still involves sending patch-sets via email, instead
> of just saying:
> "Here's a new feature that does <something...>. You can fetch
> it from <publicly accessible git repo/tag>, which is based on
> <upstream version>."

Hello

Some reasons in support of submission via the mailing list:

- avoids the potential inconvenience of setting up a publicly
accessible repository;
- easier to inspect than interfacing with multiple different
repositories for small changes;
- trivial to process offline for those who already read their mail offline;
- ensures a more permanent record of submissions than any privately
controlled repository could;
- …

> Is the issue that people shouldn't be forced to use git? Are
> the existing tools for dealing with emailed patches really
> better and if so, what are they?

Git is still very much used, by most contributors anyway.  Excluding
the reasons mentioned above, there is not really any convenience lost
or gained by using email- versus repository-submission.

The emailed patch sets are generated by git-format-patch(1), and
subsequently merged with git-am(1).  These patch sets contain the full
details of the relevant commits and the result is no different to
using git-merge(1) instead.

Regards