nmbd does not start without smbd (fwd)

[Luke Kenneth Casson Leighton]

> > sure, wthis could be deleted but only if usernames of, say
> > "TRUST_ACCOUNT$" are never observed except if the response to these is
> > expected to be the same.
> 
> I've no idea what you are talking about! I can't even parse that
> sentence :)

ok, let me try again.  when a workstation contacts a pdc with a udp packet
like this, and the pdc is a trusted domain controller not the pdc for
that workstation, the trusted dc sends out a response, and from memory
(last time looked at these, 6-8 months ago) it contains a SID.

i have vague recollection that there is also a mechanism to look up a
trust account (TRUST_ACCOUNT$) i.e workstation, server or inter-domain.

 
> > i _think_ the SAMLOGON response expects to contain the sid
> 
> umm
> 
> > (ok, there's _one_ of these UDPs that expects a SID)
> 
> really? which one?
> 
> our code doesn't send a SID in response to any of these, so if one of
> them does expect a SID then we must have a bug.
> 
> one of them does _receive_ a SID (which we currently ignore without
> any apparent ill effect) but none _send_ a SID.

ah.
 
> so, do we delete that code?


methinks not yet.  does anyone want to generate some UDP SAMLOGON, GETDC,
SAMQUERY packets, print them to a file (if ysing netmon) and send them
here?  the more the merrier, text-only for preference.

> > i checked the cvsweb diffs (all of them) and it's been there all the time.
> 
> you're right. I mustn't have ever committed it, or I lost the change
> in the great tea trolley disaster of '98.

sounds bad.  did someone put arvidsjaur on a tea trolly and roll it out
the window or something?