[jcifs] RAPs, RPCs, and API Design

[Christopher R. Hertel]

I don't want to hang on too tightly to the past (any more than Microsoft
does) but I do believe that support for legacy CIFS features is important.  
That includes support for NBT, RAP, mailslots, etc.  All current versions
of Windows still support this legacy functionality.

I also know, from my experience at "the office", that there are still a 
lot of folks who are reluctant to give up their Win9x & ME systems.  In 
addition, Windows XP-Home still relies on "old style" Windows networking.

Creating a RAP class (that can be loaded 'optionally') would be a nice bit
of leverage, and it would be good to just check it off of our 'to do'
list.

Just my 2cents.

Chris -)-----

On Fri, Feb 06, 2004 at 03:40:20AM -0500, Michael B Allen wrote:
> eglass1 at comcast.net said:
> > Mike,
> >
> > Is 0.8.0 "feature frozen"?
> 
> Sort of, but if it doesn't interfere with the core too much I would
> incorporate new stuff.
> 
> > I've been tinkering with some stuff that might
> > be interesting, but if I recall correctly 0.8.0 was a cleanup-oriented
> > release.
> 
> Actually this was the "everything that people would expect jCIFS to have
> but it fell through the cracks" release. I was going to do another release
> for cleaning out all of the stuff I believe I meantioned not too long ago.
> 
> However my time has been limited lately and I've been contemplating
> pushing the cleanup into this release. It seems pretty stable. At least
> the problems are pretty simple.
> 
> >  Specifically, I've got a general RAP call framework and a big
> > chunk of RAP calls implemented (using Samba's clirap2.c as a reference).
> > This allows you to create/delete users and domain groups, as well as
> > enumerating users and groups.  Not tested very well, but useful ;)  I'll
> > send it to you directly if you like (probably too big for the list).
> 
> There are a few problems with this. First, I'm still trying to catch up to
> your last contribution. As I've said my time is slim, so I would much
> rather work on your RPC code if I'm going to do Windows management
> functions. RAP is the client side of deprecated Windows operating systems
> (although Windows 98 was resurrected at the last minute I believe).
> Second, the RAPs aren't hard, its the APIs that need a lot of thought.
> When we expose the idea of users and groups it has to dovetail with ACLs
> which needs to consider permissions etc. Theres a lot of abstract
> brainstorming there.
> 
>   [maybe someone on the list has some isightful input here?]
> 
> If you want to think about these APIs, explore the object model, and then
> create an implementation that uses your RAPs then that would be
> interesting. The implementation wouldn't have to be really solid. We just
> want to isolate how the concept of users, groups, and permissions exposes
> itself through the jCIFS APIs in anticipation of getting RPCs going.
> That's really moving forward to a common end (RPCs).
> 
> I think the API design would also be a fun thing to do. You would be good
> at it too because you understand the Windows security model well which
> would be elemental to the overall design.
> 
> But by all means send me the code. I'll put it in the download area for now.
> 
> Mike

-- 
"Implementing CIFS - the Common Internet FileSystem" ISBN: 013047116X
Samba Team -- http://www.samba.org/     -)-----   Christopher R. Hertel
jCIFS Team -- http://jcifs.samba.org/   -)-----   ubiqx development, uninq.
ubiqx Team -- http://www.ubiqx.org/     -)-----   crh at ubiqx.mn.org
OnLineBook -- http://ubiqx.org/cifs/    -)-----   crh at ubiqx.org