anyone thinking about extending bayes classification to complex numbers? it seems like classifying two-dimensional fields might prove interesting...

i've been working on kchuid again, and i'm starting to get excited about the prospect of having a setuid-less machine someday soon.

i added chdir/chroot support; that means i can change the working directory and "root" of another process. kind of like yanking the carpet out from underneith them... i was actually kind of suprised it seems to be working...

like everyone else, i saw that bayes application on spam filtering on slashdot recently. it seems to me that this could easily be extended into a rate-voting system. and then i noticed avogato actually already does this... only it's people certifying people instead of people certifying spam.

i'd like to see a wide deployment of a system like this... my mailbox is getting full...

well i did it...

i wrote a ldap-client that doesn't depend on openldap... still need to add kerberos support and regress my filters, but it should be enouh to add to my nameserver and finally end some of the worst outstanding problems.

it will be glorious

another version... another day. my week left me with some good and some bad. good: distribution, bad: openldap. i almost bit the bullet and tried to remove dependencies on openldap, and write a full server.

i'm kind of glad I didn't... more on that later...