Date: Wed, 9 Jan 91 12:06:54 -0500
From: John T Kohl <jtkohl@MIT.EDU>
Meeting summary from 1/8/91:
tytso: kprop/kpropd are real close, sendauth/recvauth are done
currently in debugging wait state
jon: hammer is close; the program structure is all set, but
something in the encryption/decryption isn't working
right; in addition multi-component names aren't working
right (but they do work with kinit &
no admin server progress
jtkohl: draft4 KRFC is out; will be working on EUUG/EurOpen paper
has been doing some small library tweaks
jfc: DEC has received and mostly agreed with his comments on
the GSSAPI C binding; he will be receiving a new
user2user: has some preliminary code, but wants to
understand issues & such in more detail before
circulating document for review
Discussion of KDC database implementation of string2key seeding; end
result is a new field in the database with length/contents; the length
is a 24-bit bitfield, with 8 bits left for a type tag (to encode common
cases); the contents will be empty for those common cases.
Discussion of API w.r.t. allocating/filling in/freeing structures for
functions which need to return structures via the parameter list.
Consensus is that uniformity of interface and existing malloc/free
overhead make it worthwhile to require all such API entry points to take
foo_struct **, and fill in the pointer to point to a newly-allocated
*** as time permits, each team member should identify such API functions
and note them in the krb5dev meeting; Ted will work on converting them
when his kprop/kpropd work is finished.
/mit/krb5/src/TODO2 contains a list of known outstanding
coding/documentation work with (names) indicating who's known to be
working on them.
some discussion of random number generation algorithms & seeding
functions, and what Kerberos should use (the KDC uses a DES-based RNG;
sequence numbers in general need only be sparsely distributed so they're
not re-used under the same key/in the same session).
*** Next meeting is in E40-316 on Thurs 17 Jan 1991, 3:30pm