Subject: Re: LKM support
To: None <email@example.com>
From: Ty Sarna <firstname.lastname@example.org>
Date: 11/11/1996 13:00:44
John S. Dyson wrote:
> ^^^^ You are misinterpreting my position.
I certainly simplified and overstated it for effect, yes :-)
> My initial (read: simple) mods would be to MFS (please read my message.)
> Alot of bang for the buck can be gotten by some mods that would
> cost minimal amounts of time. (Est 1-2 days of work for me.)
Nothing wrong with that.
> It still needs the naming structure, etc... I didn't think anyone
> would be interested in developing another set of directory/file semantics.
Implementation != semantics. But in fact, not all of the semantics are
necessarily desirable (the value of the immutable and append-only flags
seems less clear, for example. There are probably better examples.)
> IMO, one of the mistakes in the MSDOSFS is that code wasn't derived using UFS.
Perhaps, but it too is a disk-based filesystem.
> The MFS "problem" can be solved in various ways. So, we disagree more
> than I had thought, because it just seems to me to be a waste to
> reimplement userland filesystem semantics (and I assumed that we would
> naturally agree.)
Well, it depends on how it's done really. There may be value from some
parts of ufs, however, I suggest a design still be done without any
though of UFS, to stay pure. If when the design is done there is overlap
with UFS, use it for the implementation. I just think the circumstances
of a ram filesystem are different enough than anything we have now that
it really deserves a truly fresh approach.