[3265] in linux-net channel archive
Re: Bug in 1.2.13 firewall?
daemon@ATHENA.MIT.EDU (Dennis)
Mon Jun 17 14:41:35 1996
Date: Mon, 17 Jun 1996 11:59:13 -0400
To: David C Niemi <niemi@wauug.erols.com>
From: dennis@etinc.com (Dennis)
Cc: linux-net@vger.rutgers.edu
>On Sun, 16 Jun 1996, dennis wrote:
>...
>> The idea that a commercial vendor would invest corporate resources
>> in a value-added software driver and give away source so that their
>> competitors could use it is simply not practical. Perhaps for a bare
>> bones or highly board-specific driver...but most of the "value-added"
>> features that make our product attractive are portable to other products.
>> So what you're saying is that you only want raw board drivers, and you
>> guys want to write the frame relay and do all the value-added stuff
>> yourselves. The result of that is that the entire community has to wait
>> much longer for the features, and they are stuck with a single set of
>> features for the entire O/S.
>
>There is another possibility, which may be a good idea licensing issues
>aside. Write a generic minimal kernel driver for your card, and do all
>the really neat proprietary stuff in user mode. This can insulate most of
>your code from the vagaries of kernel changes, and in fact most likely
>from the differences between several UN*X-ish OSes as well. Whether this
>makes any sense for your driver, you be the judge, but it is often a good
>idea.
A better idea is an object library that gets linked with the minimal source
driver.
Userland solutions are not good performers at high speeds...our product is
optimized
for kernel-level access.
We've done this before....but not in a minimal fasion. We decided to stop
providing
source because no-one was using it and it was a big hassle to maintain and we
wanted to lower our prices. If someone will commit to taking the minimal driver
and integrating it with the LINUX kernel then we may consider it. Boards for
such
a driver could be sold for less as well.
Dennis
>
>
>> America thrives on competition (and I recognize that you're not from here),
>> and there isn't enough difference between decent hardware to get
>> people to buy your product without a price war. If we made source available,
>> the first thing that would happen is that someone would port
>> the features to a less expensive board. So then we have to drop our
>> prices (which you THINK is a good thing), cut back on support and then
>> we stop doing new development for LINUX and start concentrating
>> on NT or something where we can make better margins. The effect
>> is that the LINUX community loses.
>
>If most of your added value is in your software, of course you have to
>worry about this. But running to NT does not help if there are
>lower-priced cards out there; you will doubtless have competition wherever
>you go. If you are saying that your hardware is inherently more expensive
>than other hardware that does the same thing, I think you are in trouble
>regardless...
>
>
>> Another ramification of the "source" distribution is that it becomes
>> impossible
>> to support software that has been ported by a user. We are commited to
>> support the product, but if all of our users are running different
versions of
>> our software (even if its just recompiled) it damages the entire process.
>
>I think that is an unnecessarily extreme attitude. Just because you make
>source available does not mean that you cannot make an "official" binary
>version available for a couple of key "stable" kernels. If someone
>reports unusual problems, ask them to try an "official" version before you
>spend any effort on it (or even state that any unofficial versions are
>unsupported, if it is so much of a problem).
>
>
>> The best one that I heard was the "let 'em put it in E-PROM". Tell me,
>> why is that acceptable? You want to pay $100 shipping and handling
>> every time there's an upgrade, or do you want to download from an
>> ftp site? Another thoughtful quote from the peanut gallery.
>
>A joke, I hope.
>
>---
>
>There are undoubtedly some people who will take the spirit of free
>software so far as to hurt their own cause. I hope you are not too
>annoyed to continue supporting Linux, whether with binaries or source.
>While a source distribution is more convenient for those running at newer
>kernel revs, or those able to do their own kernel debugging, I think that
>distinction will be a bit less painful for a few months now that 2.0 is
>out and you support it.
>
>With a little reflection the more zealous Linux boosters should also
>realize that vendor support for Linux is far more important at this point
>than quibbles over how that support is achieved. Linux is still a little
>fish, and the Linux community cannot afford to scare away commercial
>support or fragment the community with infighting, or it cannot continue
>to grow. In fact it is an emphasis on code that works over ego and
>politics which has let Linux come this far this fast.
>
>Cheers,
>
>David
>Niemi@wauug.erols.com 703-810-5538 Reston, Virginia, USA
>------ Money talks, but it is wrong half of the time. -----
>
>
>
