SNIA to Release SMB/CIFS Docs

juan large moose writes: "According to this post on the Samba Technical mailing list, the Storage Network Industry Association will shortly release version 1.0 of their own SMB/CIFS technical reference. Version 0.9 has been available for some time. Last week, Microsoft released their own CIFS specification, but there was a catch. According to a statement in the document, readers were not permitted to implement software based on the Microsoft spec. unless they signed a license agreement. The SNIA document comes with no such restrictions. Several organizations, including IBM, HP, the jCIFS Team, Network Appliance, the Samba Team, etc. contributed to the SNIA document."

This document is biased crap

[PhysicsGenius]

In order to advance my supercollider work I had to read this doc and implement it in a phase-space invariant manner. One classic quote:

Under no circumstances should a SMB server support UNIX-style long names.

Hey thanks for throwing out standards and compatibility, guys!

Re:This document is biased crap

[PD]

top - 3 characters
bottom - 6 characters
up - 2 characters
down - 4 characters
strange - 7 characters
charmed - 7 characters

So what are you worried about?

Just another useless rehash

[selectspec]

The 0.9 version was just another rehash of the original microsoft rich text documents published in 92. Without the RPC/DCE API, you can't implement CIFS. Unless 1.0 includes the RPC/DCE API codes, this is just another useless document.

Re:Just another useless rehash

[juan large moose]

You need the NBT and CIFS stuff to implement the framework for carrying the MSRPC garbage. Otherwise, you're absolutely correct. Too much of the filesharing system relies on MSRPC now.

Thing is, most of what is done via MSRPC could be done without. MSRPC obfuscates things and makes the protocol more bulky, unwieldy, and annoying. I'm sure it benefits someone somehow. It seems to me that whenever the coders in Redmond want to do something they just add another RPC call to handle it...even if it can already be handled at the SMB/CIFS level.

The reason the SNIA stuff is interesting is that it is more complete and correct than the release Microsoft did last week, and it doesn't have the goofy licensing flaming hoop. That helps to show that Microsoft did not provide anything new or significant by releasing their CIFS docs.

See also: http://www.theregister.co.uk/content/53/24490.html [theregister.co.uk] .

Personally, I think that the reason that there is no value in Microsoft (or anyone else) releasing CIFS docs is that the protocol has spun out of control and is of no value any more. It needs to be retired and replaced with something that is open and freely implementable. Not a likely scenario...

-- JLM