I'm not so sure about this anymore. I wrote that code back in 2007 and it did an OK job on the Unix type of requests.
However, a good solution for the Mac requires a lot of unusual things to work, e.g. the extended attributes and the 
little known volfs pseudo file system.

The userland only solution was built by intercepting the system calls from the application before entering the kernel and then simulating in userspace what
the kernel would have done in case there had been a real kernel module for coda.

Now that the Lion has entered the scene, I'm not sure this approach will work as at least some applications are now run in some kind of a sandbox
where they have to declare beforehand what resources they will need. 
I haven't tested anything, but this could very likely make it hard to do what the userland-only solution did. 

-- Christer

16 dec 2011 kl. 15:43 skrev <u-codalist-s7mc_at_aetey.se> <u-codalist-s7mc_at_aetey.se>:

> Hi Piotr,
> 
> On Fri, Dec 16, 2011 at 12:13:35PM +0100, Piotr Isajew wrote:
>> I'm going to get back to using coda after about 10-years
>> break :) I'm specially interested in accessing it from Mac OS X
>> 10.6 clients. Could someone update me about the current status of
>> this? 
> 
> Your best bet would be fixing Userland-only Coda which worked
> but was never reached a release.
> 
>> I googled around for a while and it seems that DarwinCoda project
>> is dead. I found a mirror (?) for it's web page but have no idea
>> where to get sources from.
> 
> Sources are available but Apple has changed a lot in the kernel since the
> time when that code was suitable. If you are interested in reviving that
> effort, we can try to ask the author if he can give you any extra advices.
> 
>> Last post (~2007) on this list mentions that kernel extension
>> needs to be ported from 10.3 to higher versions. Is that still
>> true?
> 
> Yes. I'd suggest rather fix/use the userland solution.
> 
> Regards,
> Rune
>