The poll was a user that apparently wanted to re-write in java, which as far as I'm concerned it completely and utterly pointless. A direct port from to java would gain nothing, and building a new system that borrows from scoop would no longer be scoop. There are plenty of problems with the system, but they're being dealt with piece by piece.

Right now, blocks and vars still need a little more seperating (namely, some vars still control colors/fonts/etc. which is what blocks need to). Once these are split apart (which shouldn't be too difficult, considering the calling of blocks is the same as vars). Also, all the block programs should probably be moved over to vars, with an additional var type added to give textareas. Otherwise, though, the difference is pretty clear: blocks control display, vars control logic.

I don't really know what you mean by too many clicks, though in general I'm sure the interface could use some changes. Also, the HTML that's currently hard-coded into the code need to be moved into blocks (this is an on-going project).

SQL caching is something that was attempted and failed largely because the cache grew too quickly. While that part's been fixed, it's something that probably won't happen because it's entirely un-necessary. Caching is better done at the application level, as is already done. SQL caching still has the over-head of parsing and formatting the data, while caching at a lower level allows for finer-grained control.

Chatterbox and in-site messaging are both things that people have been working on, but I'm not sure where they are right now. I know hillct was working on a chatterbox that can also do messaging, but I don't think he ever finished. Right now, he's working on XML-RPC support, which will make it easier to do things such as distributed login and other types of integration.

Already it's entirely possible to split scoop up across servers, as k5 already does this. It runs a db server, three or four mod_perl servers running scoop, two proxy servers on the front-end which serve static content and load-balance to the backend servers. The only thing scoop can't do right now is read from a replicated db, but that's not a major change, and I'll be sure to include it with the db interface changes I'm working on.