Mini C project - Digging the source ?

Richard Sharpe sharpe at
Mon Jan 7 13:58:01 GMT 2002

Christopher R. Hertel wrote:

>I am very enthusiastic regarding libsmbclient.  It is a very good idea and
>one which I am very glad to see happening.  However, I have a few issues
>with the design of libsmbclient.  These are not problems, per. se., but in
>specific cases can cause trouble. 
>Libsmbclient is based on Samba and uses a lot of samba-isms.  On any
>system with nmbd running, that's okay.  If I recall correctly, however,
>libsmbclient does things like read packets out of the unexpected packet
>database.  Again, this is okay on systems running nmbd.
Actually, I never did that, but did similarly egregious things. It no 
longer does anything like that.

>Also, libsmbclient is designed for Unix.  I know Ulf's platform, and it
>isn't Unix.
Yes, that is true :-)

>Personally, I'm working on an NBTD.  Several people have asked about this. 
>The idea is fairly simple:  Run a daemon that provides the virtual NetBIOS
>layer *separate from* the other layers.  This has some disadvantages.  For
>example, one reason Samba is fast is that we can build and parse packets
>all at once.  Dividing up the layers would slow things a bit.
>On the plus side, it makes it easy to switch between port 139 and port 
>445 semantics.  It also makes it easier to write light-weight clients 
>and servers--the equivalent of W/9x peer-to-peer type stuff.
>Finally, jCIFS already does a lot of what Ulf is looking to do.  Ulf, if 
>you can run Java on the Amiga or convert Java to C or C++ you may be in 

Richard Sharpe, rsharpe at, LPIC-1,, SAMS Teach Yourself Samba 
in 24 Hours, Special Edition, Using Samba

More information about the samba-technical mailing list