The Tux3 filesystem project is restarted

Daniel Phillips phillips at
Fri Jun 8 20:13:24 PDT 2012


After laying dormant for two years or so, the Tux3 filesystem project is now 
officially restarted, thanks mainly to Hirofumi, both for believing in the 
design and tirelessly committing a high volume of quality code over the last 
few months. By now, I think that Hirofumi may be responsible for more lines of 
Tux3 code than me, including all the kernel-specific code and a high quality 
test harness that puts to shame my own efforts in that direction.

As of today, atomic commit is working in user space and not far from working 
in kernel. Some three years ago, I said that a properly working atomic commit 
would be the trigger for submitting the Tux3 driver to mainline kernel, and 
now that does not seem far away. But of course we will take our time and get 
it right.

We are continuing with Tux3 because we believe we have something substantial 
to contribute to the modern science of filesystem design, and that the basic 
design premises will result in an efficient, full featured filesystem that 
performs well and is expressed as a tight, highly maintainable code base.

Besides Hirofumi's voluminous contributions, there is some very nice work on 
block deduplication from the student team at Pune Institute, now no longer 
students but working professionals in the technology industry. This needs to 
be further developed and merged. And there are other great contributions, such 
as the original Fuse prototype from Tero Roponen, which has been improved by a 
number of artists to the point where it is nearly usable as a filesystem in its 
own right.

So: greetings and salutations to all Tux3 contributers and supporters, and 
please know that your considerable effort has not gone to waste.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Tux3 mailing list