8 Oct 2001
(updated 9 Oct 2001 at 00:17 UTC) »
well, the site goes well: behind the scenes is some
work going on. osexchange moved to dcerpc.net. NTLMSSP as
an auth/sign/seal module, which is like the first real step
in getting NT interoperability, is slowly progressing well.
i have a successful server-side authentication and sign/seal
decryption working, as long as the password is "test".
the problem here is the dependencies. NT authentication
requires NT services and an NT-compatible API. without the
NTLMSSP in place, i can't test the authentication because i
don't have an
API to use to actually validate the user!
_fortunately_ i can use rpcclient which i developed over
four years as a test-tool. this i am really pleased about,
and am learning a lot more - having access to freedce source
- about dce/rpc than i was by just bouncing packets off of
what i am really pleased about is to have a third
dce/rpc-compatible "thing" in the picture against which to
_un_fortunately, this finds that there are bugs in the
entire TNG/samba rpcclient / samrd NDR marshalling /
unmarshalling libraries. well, of _course_ they are. it
looks like microsoft was being incredibly lazy and just
getting "wire-compatibility" [just like me *grin* except i
have an excuse: for copyright and interoperability reasons,
i couldn't look at any specifications]. so, if it "worked"
when we bounced packets off of NT services, well, then, it
worked, and that's the end of it.
but now we have _real_ functionality, by the people who
actually _wrote_ dce/rpc, and of course, rpcclient fails in
specific instances against freedce because we didn't know
gonna be interesting.
and a lot of work, for which i am still not being paid.
which is irritating, to say the least, given that this
effort provides the entire open source community with an
extremely valuable tool which, if nothing else, finally
allows them to catch up with microsoft's last ten [physical,
not man] years of development in key, strategic areas.
number of development man-hours for TNG / dcerpc.net
CRITICAL PATH components comes to over 600 man-hours.