[973] in linux-net channel archive
Re: plip: support more hw (patch inside)
daemon@ATHENA.MIT.EDU (NIIBE Yutaka)
Wed Aug 23 20:21:48 1995
Date: Wed, 23 Aug 1995 12:33:44 +0900
From: NIIBE Yutaka <gniibe@mri.co.jp>
To: Philip Blundell <pjb27@cam.ac.uk>
Cc: Alessandro Rubini <rubini@ipvvis.UNIPV.IT>, linux-net@vger.rutgers.edu
In-Reply-To: <Pine.SOL.3.91.950822140607.14711A-100000@hammer.thor.cam.ac.uk>
I changed Cc: to `linux-net' from `linux-kernel', as this is
networking issue.
On Mon, 21 Aug 1995, NIIBE Yutaka wrote:
> If there is enough reason to support 2-bit protocol, we'll include the
> feature. However, your current implementation makes PLIP pretty
Philip Blundell writes:
> Are you sure that 2-bit PLIP is a worthwhile idea? I have a feeling that
> it will turn out to be *slower* (certainly not much faster) than SLIP/PPP;
> far more processor effort is needed, and it's not full duplex.
No, I'm not sure that 2-bit PLIP is worth to be implemented and/or
worth to use. I don't have enough information so far. Currently,
I'm skeptical.
First, I don't know if new "optimizing" (or broken?) parrallel port is
popular or not. (I think that it doesn't conform centronics
interface.) There are other software which use parrallel port like
4-bit PLIP, e.g. Interlink, Purdue's PAPER, however, I don't know
such usage like 2-bit PLIP.
Second, we don't know how fast 2-bit PLIP is yet. The one which
Alessandro has posted is his first implementation. Hopefully, it may
be improved more. I think it is better to continue hacking for both
of speed and stability.
I think, it is not good idea to integrate 2-bit PLIP and 4-bit PLIP
into the same module and changing the mode on the fly. It may
convinient for users somewhat, it makes PLIP slower. Alessandro, why
don't you try to separate 2-bit only PLIP?
--
NIIBE Yutaka
Mitsubishi Research Institute, Inc.
