CVS update: samba/source/include

[Luke Kenneth Casson Leighton]

On Mon, 20 Dec 1999, Jeremy Allison wrote:

> Luke Leighton wrote:
> > 
> > Date:   Monday December 20, 1999 @ 11:36
> > Author: lkcl
> > 
> > Update of /data/cvs/samba/source/include
> > In directory samba:/tmp/cvs-serv4705/include
> > 
> > Modified Files:
> >       Tag: SAMBA_TNG
> >         ntdomain.h
> > Log Message:
> > beeaaautiful.  the one thing that i hate doing on this code (third or
> > fourth time, so far) is messing with the logic to reconstruct
> > request and response data streams from multiple DCE/RPC PDUs from multiple
> > SMB requests and responses.
> > 
> > i am therefore very pleased to announce that the second round of mods,
> > this time to get "long" DCE/RPC responses (across multiple PDUs) is
> > operational.
> > 
> > all that is left is to test / debug "encrypted" RPCs, and _then_ i am
> > in a position to add the (18-month-awaited) "long" DCE/RPC requests,
> > which neither old nor new versions of samba code currently support.
> 
> Hang on a minute Luke - doesn't the RPC rewrite in 2.0.x
> do "long" requests correctly (ie. merge them into a buffer
> on request/response).

responses, yes.  requests?  i doubt it.  i didn't code that, i don't
understand it fully...  looking at rpc_client/cli_pipe.c:rpc_api_pipe: no
it does not.

examining write_to_pipe , i find that it assumes that the write will have
an msrpc header at the front of it.  this is not the case.

so, no: both client and server do not do MSRPC that spans multiple SMB
writes.

> Or is this some new definition of "long" I don't yet know
> about :-) ?

:)