From: becker@super.org (Donald J. Becker) Subject: Re: Why not BSD-FFS? Was: [Q] newextfs Date: Thu, 11 Mar 1993 04:12:44 GMT
In article <1nlneoINN860@senator-bedfellow.MIT.EDU> tls@athena.mit.edu (Thor Lancelot Simon) writes:
>means! But I also think that the Linux community has a truly destructive
>tendency to assume that different _does_ equal better, and throw its work into
>fifty million things at once, many of which are often just reimplementations of
>something which a little bit of research in the literature might have shown
>could be done better. Sorry if I was too loud when I said this.
Hmmm, "truly destructive"? I haven't noticed many destructive
attributes of Linux community, except flames in a few threads, often
by people that haven't written _any_ code or documentation.
Some of these "reimplementations" explore different design spaces than
the originals. My ethercard drivers for instance: if I had just
copied the drivers from BSD we might not have encountered so many initial
bugs, but we wouldn't have auto-probe, auto-IRQ, ping-pong buffers and
a few other features. Some of those features might turn out to be a
bad idea (like trying to make autoprobe and autoIRQ work with all of
the WD80x3 revisions), but some will certainly live on to make Linux
better than a copy of some other OS. Do you know of any other
ISA-based OS doing 950KB+/sec TCP rates using 8390-based ethercards?
Likewise, writing new file systems is the only way to make real
progress in a research area that was once stagnant. I'ld like to see
a half-dozen different file systems under Linux, and not have a
shoot-out until they all mature. We still have room for a compressed
file system, an extent-based file system, a log file system, striped
file systems, a buffer-cache file system (a file system that sits in
front of another FS, such as a CD-ROM or SLIP-NFS, and caches accessed
pages).
Finally, installing Linux on your machine doesn't make you a
contributer to the community. I enjoy the put-up or shut-up attitude
of the Linux developers: we tolerate no one claiming to be an "old
sage" here. The only way you can make something happen is by writing
the code. If people need it and use it, they'll build on top of it or
improve it.
So, let's go out and write some code.
-- Donald Becker becker@super.org Supercomputing Research Center 17100 Science Drive, Bowie MD 20715 301-805-7482