[libvirt] [RFC] Accepting PRs/MRs for libvirt on GitHub/GitLab

[Peter Krempa]

On Wed, Oct 16, 2019 at 17:08:20 +0100, Daniel Berrange wrote:
> On Wed, Oct 16, 2019 at 05:03:31PM +0200, Ján Tomko wrote:
> > On Wed, Oct 16, 2019 at 02:22:56PM +0200, Andrea Bolognani wrote:

[...]

> > Here it deviates from the usual mailing list workflow where the patch
> > has (in theory) a chance to be seen by all the developers.
> > 
> > But given that the requests will probably
> > a) be close to trivial
> > b) seen by a group of developers, not just one
> 
> I wouldn't expect the changes to be trivial. Current stuff
> is trivial largely because we tell people not to open merge
> requests. If we adopt use of web based review, then expect

I'd still want the message we'll put out to encourage them using e-mail.

> people to submit non-trivial patches. I would do so myself
> for example. Thus I think we must make a clean switchover
> from email to a single web based tool.

I disagree. There is nothing really appealing to me in any of the web
based frontends for git.

The user interface of them is designed to be flashy but that really
hurts usability of git. We get cool icons but in return we must pay with
always-online connection, loading bars if you click anywhere and the
general necessity to interact with the browser which requires a lot of
mousing around.

The commenting interface on individual patches is very poor given what
email allows you and in many cases it's hard to access older versions
after a pull-request is force-pushed.

Also we then get to the discussion whether to use the most popular web
based git hosting site, which is the most terrible to use of them all or
one of the less popular and less bad ones. Since this idea is sold as a
way to attract outside contributors, it might lead to sacrificing
usability for popularity ...