features
------------------------------------------------------------------
* the metasearch messes with the max_page setting. 30 from goog plus 30 from teoma equals 60 results. we can:
	- split it between available engines (complicates the meta interface)
	- make a pagination interface either with
		1) javascript
		2) storing "page 2" in the DB -- this will complicate "real" pagination 
		 (that is, going back to the engines for the next set of results), though 
		 I am going to leave that feature out for as long as possible; I suspect
		 99.44% of people, including myself, don't actually *use* "page 2" if there
		 are at least 20 results.

* pagination for the browse by keyword page. once a keyword gets over 50-100 urls (give it time, give it time) it will become unwieldly. -- hook it into the max_page preference

* need to start thinking about what happens when there are 50,000 pages in the cache. Once some smart fellow installs Dowser on a server at the library and wants 50 users to wail on it at the same time.

* should preferences be in the COOKIE instead of prefs.pickle? This dovetails into the "university install" above.

* are we looking at an "enterprise" or "multiuser" fork for Dowser? take the same parts and create a "smart search" caching proxy?

* need a good, stable, garbage-collecting daemon to get rid of empty paths, broken db references, enforce disk quota, etc.

* add a feed-reader? random grabs, search & organize, spam and archives. look at competition to get a feel.

* flesh out the tips and help system.

* an auto-summary that doesn't suck. (!!)
* better "stripper" logic -- get rid of text between script, object, iframe, embed, style tags. sgml lib is cool, but doesn't allow style-stripping.


* i18n for stemming, etc. generalized stemming lib for extracting keywords, etc.

* full text search within the cache?

* import/export -- xml, text, sql, or (say it softly), RSS?

* add support for google's dictionary helper.

* need elegant way (UI-wise) to delete entire keywords from the local cache. keyword blacklist?

* better debug system, esp on windows.

* need sort by date, size

* need intelligent grouping in results

[browser integration]
	//DONE* searchbox plugin for moz-like browsers
	* bookmark watcher: bookmark a page, and it gets automagically added to the database.




nits:
-------------------------------------------------------------------

* search.cluster5 is a hog. need to prune word list sooner.

* how can browse work with un-alphabetical languages? 
what about mixed databases? german & english & chinese & korean.

* "undo" facility for deletion?

* better typechecking on the addurl thingy.... perhaps a bulkadd url thingy -- this will be obsoleted by the "bookmark watcher"
 
* respawner! -- IS there a way to kill a thread in python?

* are pre-compiled regexps with backreferences thread-safe?