From: dsb@world.std.com (David Boyce) Subject: Re: Stabilizing Linux Date: Fri, 21 Aug 1992 17:47:37 GMT
In article <3503@key.COM> rburns@key.COM (Randy Burns) writes:
>In article <1992Aug16.014647.19241@mintaka.lcs.mit.edu>, ericy@hal.gnu.ai.mit.edu (Eric Youngdale) writes:
>> In article <Bt1u3u.3zv@world.std.com> dsb@world.std.com (David Boyce) writes:
>>
>> I think you may misunderstand the market. The "paying customers"
>> mainly want a source for linux that does not depend upon network access or a
>> modem.
>> My sense is that the CDROM manufacturers (at least the ones that I have
>> been in contact with) are not interested in a lot of release engineering.
>I expect this will change as the volume in Linux disks/CD-ROMS increases.
>Right now, there isn't much real potential volume in Linux disks/CD-ROMS
>compared to commercial products. I would expect this to change significantly
>over the next 6-12 months.
Yes, this was my point. It's certainly true that the linux-sellers
of today simply want to put all the ftp-able stuff on a disk/CD
and sell it for $30. But there's every reason to think this will change.
For one thing, each subsequent release after the first one requires
more release engineering. Thus the first may be quick and easy,
but in a year or 2 when the number of linux users has mushroomed
and linux vendor X is working on their third or fourth release,
the situation will be different.
What happens the first time someone ports a commercial
(non-freeware) application to linux and a lot of people are
using it (note that there's already talk, at least, of doing so with
Motif)? Then you want to issue a new release; don't you have to
make sure the old binary runs on the new release? Or at
least find out so you can document it? You can't tell people
to UTSL.
My prediction, and only time will tell, is that anybody selling
linux will get into these weeds real quick whether they think
they will now or not. And my argument is that c.o.l people
should not and need not get involved in compatibility/reliability
release issues. Let the vendors decide how much is needed and
how and when to do it. But note that my argument doesn't depend on my
prediction: if the vendors think it only needs $30 worth of work and
support, let them give it that much and let the market decide. My
speculation that the number will creep up is beside the point.
Also, consider Mark Williams/Coherent. They sell and support
a complete system, including documentation and the works, for $100
and presumably make a profit. By all accounts they do a very good
job, and I assume that involves a certain amount of release engineering
and testing. So it would seem to follow that another vendor
could take on the task of simply supporting, documenting,
testing and release engineering linux for <$100, given that
unlike MWC they wouldn't have to pay any actual developers.
In fact, if MWC took on linux as a sideline and sold it
for $100 with their fabled support and documentation, I'd
buy it and recommend it in a second.
>> Theoretically, someone may do the release engineering, and then
>> try and charge 500$ for the disc. They would certainly be allowed to under
>> the GPL, but they would have to compete with the 30$ disk.
See the above.
This is off the track a bit, but I think people in this group
tend to underestimate the "middle group" of users. The statement
has been made that "Linux is by hackers, for hackers" and the
perception seems to be that the world is divided into kernel
hackers and unix-naive DOS refugees. But there is a large
constituency of people who are seasoned Unix users/programmers
but not kernel hackers. If, for instance, we simply look at the
group of people who (a) work for a company that has *ix computers,
(b) likes/needs to dial in from home, and (c) has a *86
machine at home for that purpose, I think right there you
have a very large number of people who would be much happier
being able to run linux with X and, say, minicom, because that's
what they use at work, than to have to use DOS/procomm. I believe
ProComm+ alone costs in the $80-$100 range and other comm packages
up to $170, so even if commercial linux cost $100 money wouldn't be
a factor for these people.
-- David Boyce dsb@world.std.com 617-576-1540