Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 6 May 1997 22:35:15 -0500
From:      Tim Tsai <tim@futuresouth.com>
To:        Michael Smith <msmith@atrad.adelaide.edu.au>
Cc:        Jaye Mathisen <mrcpu@cdsnet.net>, dennis@etinc.com, tim@futuresouth.com, hackers@FreeBSD.ORG
Subject:   Re: if_de.c ????
Message-ID:  <19970506223515.39053@shell.futuresouth.com>
In-Reply-To: <199705070108.KAA15806@genesis.atrad.adelaide.edu.au>; from Michael Smith on Wed, May 07, 1997 at 10:38:15AM %2B0930
References:  <Pine.NEB.3.95.970506130427.1963J-100000@mail.cdsnet.net> <199705070108.KAA15806@genesis.atrad.adelaide.edu.au>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, May 07, 1997 at 10:38:15AM +0930, Michael Smith wrote:
>It is interesting to note amongst all this bitching and whining about
>the "poor" support for 'de' cards that there has been almost nobody
>offering any help or support.  It's all been "where's my &^%$&^$
>de driver right *&^%*&^%*& now".  

  Michael, I don't think this is fair.  If you look at the mailing list
archives I have publicly offered to pay somebody to support 21140-AC based
cards.  I also sent mail directly to one of the core members and never
got a reply.  Also, on any system that wishes to have mass end users (and
I presume this is one of FreeBSD's goals), there are users that will not
have much, if anything, to contribute back.  Ultimately, it's how well
you support these users that determines the success.

Now let's look at the paths I've taken in the past four months:

  - Immediately deciding on FreeBSD, I scouraged through the mailing lists
and the web page and found that DEC based NIC's are reliable and great
price for the performance.  So we ordered half a dozen or so cards.
  - Installed 2.1.6.  Found that these network cards doesn't work.  Read
on the maling list that this will be supported soon, so we ordered a
bunch of NE2000 clones to tie us over until the support is added.
  - Overtime, we've upgraded to 2.1.7, 2.2-GAMMA, 2.2-RELEASE, 2.2-STABLE,
and now 3.0 on some of the machines.  The situation is the same each time.
The NetBSD de code was not incorporated on each release.  Between tracking
the NetBSD changes and the FreeBSD releases breaking the NetBSD code each
time, we've never successfully run the de code on all our machines.  I
also never felt confident that this code would work reliably and
correctly (hey, I am no kernel hacker), so I always ended up with the
trusty old NE2000.
  - Four months later, I am finding out that the de cards will never
perform as well as the Intel cards, and not as well supported either.
 
  Am I bitter about this?  Absolutely not.  Sometimes very frustrating,
yes!  We take this as part of the experience.  I am sure the core members
had intended to support de all along, but one thing or another happened
to prevent this from happening.  I can understand that - shit happens.
I don't expect any better from any other OS.  The only time I've seriously
considered dropping FreeBSD (and did, on our backup machine) was because
the SCSI changer support keep breaking between releases.  Now that FreeBSD
2.2/3.X supports chio directly, we are all FreeBSD again.  We've also
decided to forget about the de cards and just buy Intel boards.  I
understand that work is in progress or has completed on the de port, but
I don't think it matters given the positions DG and some others has already
taken.  *I* want the best supported hardware.  If that means dumping the
de cards and buy Intel's so be it.  They're cheap enough for me to be
generous.  Now, if somebody tells me that our 4G hard drives or Pentium
Pro CPU's are not very well supported...  :-)

> > And so I'll start purchasing intel cards, and 8 months from now, the
> > Novell NE2000 cards will be the hot card to have.
> 
> Or, you can help with the 'de' cards and keep your inventory.

  I really don't see why anybody would do this, when DG and others have
said that Intel boards are the way to go.  It'd be nice if somebody simply
put on the hardware list, that 21140-AC cards are not well supported and
should be avoided.  It would prevent a lot of headaches.

  Sorry to be so long winded.  I think the FreeBSD developers have done
a fine job and have done what they could with the limited resources.  That
said, it all depends on where FreeBSD wants to go.  I know that a lot of
people, given our own experience, would have given up and gone to
Linux/etc. already.  Maybe that's what the FreeBSD community wants.  Maybe
not.  You tell me.

  Tim, IMHO, of course.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970506223515.39053>