[openbeos] File system
- From: Clay Vincent Schentrup <cvs@xxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Sat, 03 Aug 2002 19:33:53 -0500
This is just a general question about the nature of the file system plans.
The way I see it, as long as BeOS API calls to save and access files don't
change, we can really design the file system however we want it--right? So
long as it's inner workings are transparent to the current API
implementation...
On that note, what precisely can be improved upon from the current BeFS.
From my understanding, XFS and ReiserFS are "faster". However I don't know how
intimately tied the file attributes are with the file system implementation,
and so this speed might not really be much of an improvement (I'm a BIG fan of
the current MIME/"dynamic attributes" system in BeOS. It seems logical
however, that given a given general model for what the file system must do,
fs's can only advance so far. I mean, QuickSort hasn't changed in decades; so
it seems there should be a very clear picture of exactly how an extremely fast,
modern file system ought to run. If anyone would care to discuss ways in which
the current FS could be improved (again so long as it's transparent to the
system) I would love to hear about it.
The main reason this is so appealing to me, is that file system and
networking are both major points that people talk about when discussing OS's.
I really want to see OBOS (can we please name this thing so I don't have to
keep typing that) be technically respectable, because I think that will have a
huge affect on its adoption, and how many geeks will choose it vs. say linux,
when they choose to go to an alternative OS. That's my morning schpiel.
Clay
Other related posts:
