Re: experimental proposal

On Wed, Nov 12, 2003 at 09:07:37PM +0100, Juliusz Chroboczek wrote:
>DD> My plan is to do some experiments to see if a so-far unused mechanism
>DD> that the X11 protocol creators provided would provide a better way of
>DD> achieving than adding on a bunch of ad-hoc extensions or creating
>DD> unnecessary incompatibilities.  The mechanism I'm referring to is the
>DD> protocol minor version number.
>
>David,
>
>Unless I'm missing something, it's a horrible idea.
>
>Using the version number does have a significant drawback: extensions
>live in a rich name space, while protocol versions live in a linear
>numeric one.  While nothing prevents XFree86 from having its
>extensions and, say, Sun theirs (and of course XFree86 implementing
>the more reasonable among Sun's extensions), bumping the minor
>protocol version requires global consensus.

Solving core protocol deficiencies in any meaningful way requires
consensus.

>checking the server vendor string.  Icky.
>
>The obvious reason why somebody would want to bump the minor version
>is marketing (we're a more recent version than you), but that's
>obviously not a reason that we (XFree86) will ever consider.
>
>The only other reason I can see is that an extension is not allowed to
>change the semantics of core requests.  However, that is not something

Actually they can (e.g., the XKEYBOARD extension).

>we're willing to do, especially so as nothing prevents an extension
>from mirroring the core requests that it needs to change in its
>private name space.
>
>Unless I'm missing something, the drawbacks of your proposition are
>sufficiently important, and the advantages sufficiently meager, to
>make it a pretty bad idea.

I'm aware of the issues, but I like to experiment, so humour me :-)

David
-- 
David Dawes
developer/release engineer                      The XFree86 Project
www.XFree86.org/~dawes