FIX: NT sending null username effects %U expansion

[thwartedefforts at wonky.org]

On Wed, 28 October 1998, Luke Kenneth Casson Leighton wrote:
> kit's not a bug in nt4: it's just that we have different processes for
> each connection.  the _first_ connection, in one smbd process, must jave
> the username etc.  the _second_ connection does not, because that's in a
> separate smbd process.

Bug, different operation than NT 3.51 -- whatever. :)

Are you saying that the second "process" is a different UNIX process?  My tests show otherwise.  From what I can gather, this is all happening over the same TCP connection to the same smbd process.  I am specificly talking about multiple requests, not multiple network connections.  smbstatus, netstat and ps show that multiple SMB connections to different shares use the same TCP connection and smbd process.

If you are using the words connection and process in a SMB call way, then I agree that the second request sent by the client shouldn't require it to send a username over a previously validation connection, but that shouldn't make the server not use a username or implictly use the guest user.

I don't care what the client is sending for a username/password or how predictably it's sending it (fighting for MS to produce predictable software is a loosing battle), but I do care that samba interpret what it receives predictably, thus my suggestion for new parameters at the end of that message.

Andy.