wayland-protocols scope and governance

[Drew DeVault]

On 2019-02-21  4:00 PM, Pekka Paalanen wrote:
> Let's forget about the prefixes or namespaces indicating anything about
> endorsement or acceptance.

I don't think using prefixes/namespaces for acceptance/blessedness is
going to be a good idea, but I do think defining some namespaces and a
scope for the protocols in them is useful. We'll want a catch-all
namespace (I like Daniel's "ext" suggestion) for stuff that doesn't fit
nicely to avoid the political problems of a protocol with nowhere to go.

Part of the problem so far has been that the scope for each prefix is
poorly defined, so when gatekeepers appeal to that scope it has been
frustrating in the past, especially without any other namespace for a
protocol looking for a home to go. Fixing those two issues would do a
lot.

> The benefit is that we avoid almost all of the design-political
> debates. The only policy question left to answer is what kind of
> extensions are eligible to wayland-protocols at all, and there I think
> we could be very lenient: anything that has a prospect of being used by
> more than one software project.

This seems reasonable, though we might as well clarify that there ought
to be a least one client and one server implementation, ideally under
the governance of different projects.

> - Look at who the maintainers for the extension are, if you know their
>   personal reputation.
> 
> - Additionally, if wanted, we could even add "reviewed-by" entries to
>   the extension meta data in addition to the maintainer entries, if
>   people want to promote their endorsement of an extension.

Are people more useful than projects in this case? I couldn't name all
of the GNOME developers and I'm sure you couldn't name all of the
wlroots developers - but a protocol endorsed by GNOME or wlroots as a
whole implies a consensus among those people and conveys practical
information about the usefulness of a protocol more efficiently.