[jcifs] Re: jarapac-0.2.0 relased

Michael B Allen mba2000 at ioplex.com
Sun Aug 15 03:43:05 GMT 2004


On Sat, 14 Aug 2004 20:19:52 -0400
Eric Glass <eric.glass at gmail.com> wrote:

> The provider thing is just address parsing; you should be able to run it
> like:
> 
> java TestNetShareEnumAll "ncacn_np:server[\\PIPE\\srvsvc]"

Actually the Makefile was missing a '..' in the classpath.

> This works for me, but gets an error parsing the NDR UUID
> ("8a885d04-1ceb-11c9-9fe8-08002b104860").  This is just because the
> timeLow field ("8a885d04") is declared as an int in UUID.java rather
> than a long (it's a 32 bit unsigned).

Right. But an int can hold 32 bits. You just need to parse it with
Long.parseLong (although personally I think Integer.parseInt should be able
to parse that).

> I tried fiddling with it a bit
> but got another error in DefaultConnection; in transmitFragment, it
> calls processOutgoing which does a setIndex on the buffer to read the
> PDU type, but never resets it (so it thinks the buffer is only 3 bytes
> long, which is what it sends over the wire).  I got it to send some
> stuff after that, but got a fault.

Yeah, the NdrBuffer is my creation and I didn't have equivalent length
support. Fixed.

The bind is going out on the wire now and it looks close but no cigar. It
looks like the fragment lengths are messed up and the UUID is garbage. Major
encoding errors :-). Boy I wish I knew more about the protocol side.

> I'm knee deep in Davenport right now but I'll have a closer look when
> I get a chance; I think it is really close to working. :)

Ok. I'll play with it some more now then and upload the latest at the end
of the night. I look forward to seeing if you can clean up my mess :-)

Mike

-- 
Greedo shoots first? Not in my Star Wars.


More information about the jcifs mailing list