pgsql-patches considered harmful

Pursuant to a conversation this evening I would like to a suggestion:

BIRT pgsql-patches should be abolished in favour of something else that
accomplishes the bandwidth-reduction aspect without the downsides.

My complaint is that -patches serves to

a) siphon off some of the most technical discussion from -hackers to somewhere
where fewer hackers read regularly leaving a lower signal-to-noise ratio on
-hackers.

b) partition the discussions in strange ways making it harder to carry on
coherent threads or check past discussions for conclusions.

c) encourages patches to sit in queues until a committer can review it rather
than have non-committers eyeballing it or even applying it locally and
using it before it's ready to be committed to HEAD.

The only defence I've heard for the existence of -patches is that it avoids
large attachments filling people's inboxes.

To that end I would suggest replacing it with a script on the mail server to
strip out attachments and replace them with a link to some place where they
can be downloaded.

This could conceivably evolve into some sort of simple patch queue system
where committers could view a list of patches and mark them when they get
rejected or committed. I'm not suggesting anything like a bug tracking system,
just a simple page should suffice.

I fear by sending this I may have just volunteered to execute it. But if it's
the case that people support my suggestion I would be happy to do so.

--
greg