ReiserFS creator arrested over missing woman

Ben Scott dragonhawk at gmail.com
Thu Oct 12 09:21:01 EDT 2006


  First, a bit of background:

  There are really two Reiser filesystems.  ReiserFS Version 3
("reiserfs" in the kernel) has been around and "stable" for years.
Reiser4 was the even more ambitious, newer project.  According to
NameSys (http:/www.namesys.com/), Reiser4 has been production-ready
for awhile, but still hasn't been accepted into the mainline kernel.

  Thus, there are two issues: Sustaining maintenance of the existing
Reiser3 code, and acceptance/improvement of the newer Reiser4.

  I found some very current info (dated 12 Oct 2006) on the Reiser4
question here:

http://www.apcstart.com/site/amills/2006/10/1811/reiser4-to-go-mainline

On 10/11/06, bmcculley at rcn.com <bmcculley at rcn.com> wrote:
>>  Fortunately, the nature of Free(dom) Software means that software
>>only dies if there isn't sufficient interest in it.  Anyone can take
>>over the code, [...]
>
> Problem is, *anyone* can do so, competence not withstanding. The community peer
> pressure may select out incompetents in the long run, but it doesn't bode well for
> stability in the short run.

  Well, code doesn't get accepted into the mainline kernel without a
lot of peer review.  Presumably a new maintainer for filesystem code
will mean even more scrutiny than usual.  From past history, it's far
more likely that development/maintenance will stagnate, and the code
eventually be removed from the mainline kernel.  So I would fear that
a lot more than instability.

  SuSE at least used to be a big Reiser3 supporter.  It was their
preferred filesystem in their installer, and I think they were funding
development at one time.  I don't know if they still do that, but if
so, they would likely be in a position to take measures to ensure
continuity.  That's likely to apply more to Reiser3 than Reiser4,
though.

On 10/12/06, Fred <puissante at lrc.puissante.com> wrote:
> So whomever inherits the code had better spend a LOT of time getting to know
> it.

  Indeed!  That's usually a big hurdle for anyone coming into a
project, but when it comes to a filesystem, mistakes are particularly
likely to cause catastrophe.

> Device drivers in general tend to be tricky, black art stuff. Doesn't have to
> be that way, but from what I've seen they usually are.

  I haven't seen the code for Resier4, but the Resier3 code that's in
the kernel seems to be well-commented and well-organized.  I'm not, by
any stretch of the imagination, a kernel/filesystem guy, but looking
at the code doesn't make me feel hopelessly lost.  Lost, yes, but with
hope.  :)

  I believe much of the development/maintenance of the existing
Resier3 code is handled by someone other than Hans Reiser -- I recall
that Hans would hand off tasks of that nature to someone else, when I
was on the mailing list years ago.  Reiser3 was considered "done" and
in maintenance mode; Hans was focusing more on Reiser4.

> and a "help find Nina" site:
>
> http://www.ninareiser.com/

  Pretty woman.  Hans Resier can't be all that smart if he let her go.  ;-)

-- Ben



More information about the gnhlug-discuss mailing list