On Mon, 14 Jan 2002 23:39:31 +1130
Mike Andrew [EMAIL PROTECTED] spewed into the bitstream:
On Mon, 14 Jan 2002 02:57, David A. Bandel wrote:
If you have something better (than devfs), I know lots of folks who
would like to hear your idea of how to do it.
We have no argument about the 'goodness' of devfs. devfs is going to
happen, because it has to.
But if there's a better way to implement it, that would be a good thing.
As it stands, it's not that it's good, bad, or indifferent, it's that it's
the _only_ way to dynamically create what you need.
I have run devfs (past tense) I agreed with it, it did not agree with
me.
Sorry to hear that. I haven't had any problems with it or I'd have
abandoned it long ago. But it works well enough I just stick with it.
I admire your ability to use it. Richard Gooch has a *lot* of
documentation to catch up on because 80% of what is there is a 1998
argument as to why devfs is needed (in preference to other
alternatives). It is scant, to non-existent, on HOW to use it.
True. I'd say the documentation is the code, but it's nearly
unintelligible to any but a kernel hacker (which I ain't).
so you need to tell whoever owns the sr_mod module that he's got to
rename is scd_mod because he's wrong -- no?
This is facetious. The point being that the ramifications of
implementing scdX in preference to srX were not thought out fully.
Redhat is not alone, unique or the leader of this new wrinkle. And, I'd
fight anyone who said the kernel must change because of *any* distro.
True. I should have put a tongue in cheek emoticon with this. But I
haven't seen any use of scd#, only of sr#. OTOH, I don't run RH or direct
derivitives (at least not direct enough to have RH's problems).
_because_ sr_mod is hardwired, _because_ many automounters hunt srX,
this new approach may die a death and everyone will revert to srX. Right
now, there is confusion everywhere about the duality of scdX /srX and
there's no magic-cure. I don't argue the author must change, I point out
the reasons why thingz iz as they iz. My view is that the dynamic
assignment of devfs will rule the day and things will revert.
Documentation is great. And you can document the use of scd# forever.
But until devfs + all the major distros implement it (and RH and a few of
its followers aren't all the major distros), it's nothing but words. And
a number of distros do use devfs (gentoo and sorcerer come to mind, so I'm
not exactly alone).
Meanwhile, any documentation should probably cover both (as much of an
annoyance as that is).
I've been using devfs since it came out. I prefer it. It may not be
perfect, but it's a damn site better than creating thousands of
useless device nodes
No contest.
--
Focus on the dream, not the competition.
-- Nemesis Racing Team motto
Internet (H323) phone: 206.28.187.30
___
Linux-users mailing list
Archives, Digests, etc at http://linux.nf/mailman/listinfo/linux-users
