This is the mail archive of the cygwin@sourceware.cygnus.com mailing list for the Cygwin project. See the Cygwin home page for more information.
[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index] [Subject Index] [Author Index] [Thread Index]

Re: Cygwin participation threshold



On Wed, Feb 24, 1999 at 09:59:16AM -0500, Steve Morris wrote:
>Christopher Faylor writes:
> > It is interesting that you felt this way at first.  I wonder if the reason
> > has anything to do with the name "Cygwin" which sounds so similar to "Cygnus".
> > 
> > The reason I am saying this is because hundreds of people have contributed to
> > the Linux project and *many* companies make money from Linux.
>
>Actually I think you've hit on a major issue. Even though Cygnus makes
>cygwin available as sourceware it is obviously a Cygnus
>product. Cygnus controls the feature set. Design decisions are made by
>Cygnus. People can contribute but Cygnus is the final arbiter on
>design decisions and even code style.

This is very similar to the way Linux was developed except that in the
Linux case Linus Torvalds was the final arbiter.  In the Cygnus case
the final arbiter was first Steve Chamberlain, then Geoffrey Noer, and
now me.  There's still one person involved in decisions but input from
external developers is both solicited and appreciated.

>With gcc it is different. Cygnus is the official maintainer but the
>perception is that Cygnus acts more as a custodian for FSF and the
>free software community. FSF owns the copyright. Redhat is another
>example. Redhat doesn't own Linux. RPM is the only significant thing
>that RedHat copyrights and even that makes people nervous.

Of course, gcc wasn't initially developed by Cygnus either.  One of the
reasons behind the formation of EGCS was that there were a number of
contributors besides Cygnus who felt that they weren't able to contribute
to GNU's gcc project.  This is entirely a different situation from Cygwin.

Although RedHat doesn't own Linux, it obviously does make money from its
sale.  This means that if you contribute some nifty code to Linux you
also might benefit RedHat.  I don't see how the ownership issue matters
here.  Somebody is still making money from your work whether it's Cygnus
or RedHat.

>On the other hand Cygwin is obviously branded. Even the mailing list
>is controlled by Cygnus. The developers mailing list access is
>restricted by Cygnus engineers. The official Cygwin web page is
>controlled by Cygnus. The bug list is an internal Cygnus system.

All this is true of EGCS, too, isn't it?

The developers mailing list is supposed to be open only to people who
are willing to contribute to the development of cygwin.  If the rest of
the (developers) mailing list thinks that that is too restrictive a goal
then I'd be happy to open it up.  We have maintained cygwin-developers
as a closed list to try to limit discussions to actual development
issues.

>Psychologically it doesn't make me feel like I would count as much as
>a Cygnus engineer if I contributed. Helping Cygnus with their free
>software product doesn't have the same cachet as helping Linus Torvald
>with his. Linus stands first among equals partners. How can I feel
>like an equal partner to a company?

Have you actually read the linux-dev-kernel?  I don't think that
net contributors would rank Linus "first among equal partners".  Linus
is a semi-benign despot.  He accepts patches from people but he always
has the right of refusing patches and he does reject patches often.

Patches from people like Alan Cox or David Miller or Stephen Tweedie are
likely to be given the "fast track" into the main development kernel.
Patches from you or me are likely to be ignored unless we can really
prove our technical competence.

For a long time, this same model was also used by Larry Wall for perl.
In fact, it didn't really break down until Larry suffered health
problems.  Now perl has a number of people contributing code to
an volunteer maintainer who acts as an arbiter of what gets in and what
doesn't.  But in case of deadlock, Larry's word always trumps everyone.

I think that this model is actually what makes these packages a
success.  They have one person at the top with a clear vision of where the
product is supposed to be going and what it is supposed to be doing.

This is what we're trying to do with Cygwin too, with, so far, limited
success.  Cygwin does have it's contributors and I'm *very* grateful for
the time that they've put into making it better.  Some of the major
Cygwin subsystems have, in fact, been contributed by people who managed
to get beyond this "company" barrier.  Or maybe it never occurred to them.

>I guess the issue is not companies making money on free software.
>Instead the issue is companies being perceived as controlling the
>software development.

Well, if that is your feeling, I can't dispute it.  If I understand what
you're saying correctly, your philosophy for Cygwin is that you will use
it and hope that it improves from release to release but, if it doesn't,
the barrier of a company judging and profiting by your code submissions
is too high for you to consider attempting any improvements yourself.

>Tcl is entering the same delicate state. With Ousterhout starting
>Scriptics which is now the official distributor of the release people
>are beginning to get nervous. The question always hovers "will
>Scriptics pull Tcl in and make it a commercial product?" TclPro is
>$1000 a seat. What if new development or the good extensions only
>appear in TclPro? Nobody begrudges Ousterhout's right to make money on
>his major contribution but still there is anxiety.

I don't know anything about Scriptics' philosophy of business but I hope
that most people on the net know that Cygnus has been dedicated to the
concept of free software for a long time.  Hopefully that stands for
something.

If it doesn't then I don't know how to overcome this obstacle.  I could
give personal assurances that Cygwin will always be free but since
Cygnus is a business, DJ and I could be booted out tomorrow and replaced
by some evil software hoarders.  It's not a likely scenario (at least
the evil software hoarder part) but there is no guarantee that anyone
can give that means anything.

But, of course, maybe none of these observations matter.  We're talking
about perceptions, here.  It's my contention that if the EGCS project
had been named 'cyg-gcc' that it probably wouldn't have as large a base
of contributors.  It's probably the initial 'cyg' which is off putting
to people as well as the technical barrier of having to know both UNIX
and Windows code.

cgf

--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com