A long-overdue update to SQL Session is out. This version (0.2.9) will be the last version in the 0.2.x series (assuming no brown paper bag bugs in this release). I will be now working on 0.3.x, which will hook into the transaction machinery so that it only makes a single set of updates in a transaction. This should lower the database hits significantly. Installation: If you have a previous version of SQLSession, you might need to go to the management panel of your Session object after installing the new version and hit 'Change'. Note that the default schema has changed to avoid bad words like 'name', 'session', 'sessions' and 'value', since some databases have issues with these. Changelog from 0.2.3: Code now uses it's own decodestring/encodestring, because the standard python lib supplies an incredibly stupid implementation. (take a string, wrap it in a StringIO, then pass it to encode which then pulls out the string again.) Rough testing shows that a small string encode or decode drops from 1ms or so to about 1/20th of that. Given how often this is done, it makes a big big difference. Jeff Hoffman supplied a patch that makes SQLSession work with 2.2 - adding __allow_access_to_unprotected_subobjects__ = 1 to the SESSION object. Ooo, magic. Changed the default schema and queries to avoid using the words 'session', 'value' or 'name'. Different databases have different ideas of which are and are not reserved words. Added a new option 'path', to specify the cookie path. This means that you can specify, say, /foo, and only the browser will only send the cookie when requesting URLs beginning with /foo. Re-did some of the internals to avoid what may or may not have been a refcount cycle. I'm still not sure what caused it - assigning the SESSION object to REQUEST.SESSION, with SESSION keeping a reference to the parent Session object. It now finds Session by acquisition. A number of dumb bugs were squished. Anthony -- Anthony Baxter <anthony@interlink.com.au> It's never too late to have a happy childhood.