httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chuck Murcko <>
Subject Re: Apache legal question
Date Tue, 07 May 1996 01:13:07 GMT
Tom Tromey liltingly intones:
> Chuck> Well, that's the crux of the matter. If Cygnus builds from
> Chuck> standard release code, making no changes other than version
> Chuck> number, then -inet should be in ServerVersion.
> Whether or not we do this depends on what you mean by "standard
> release code".  For instance, we've changed the configuration scheme
> so that Apache can fit snugly into our build process.  I personally
> wouldn't count this as a change, but it does mean that our source tree
> differs in some ways from the ones you see on the net.
Well, it sounds like two separate but relatively closely coupled source
trees, with the Cygnus tree being a little ahead of and having a few
more ongoing differences than the base Apache development one.

With things like the GNU tools, do you arrange to periodically
resync your source to the base development stuff, or do you just fetch
a new release and go from there?

> Likewise, we will probably be applying patches to our source as
> required.  For instance I've applied the mod_dir bug fix patch to our
> sources.  This means that our "1.0.5" differs in at least this way
> from the net "1.0.5".  Does this really count as a change?  If so,
> doesn't that mean we can't fix any bugs in our version of Apache?
> (The problem here is one of synchronization: we can't wait for Apache
> releases in order to do our releases)
Understood. See above. Apache folks are making an effort to do better
at timely updates, thanks to Paul and others' efforts with CVS, but
many of us (inc. me) have other jobs which bleed away our available
time. I wouldn't consider 1.0.5pl1 (although we don't use that
model here) to be a different server, and I think your question on
bug fixing leads to at least a peripheral presence on the develop-
ment side of things (this list), rather than some sort of taboo on
you doing any repairs. I'd have to leave further involvement to
the consensus of the group, but I'm of the opinion Brian, at least,
is pretty interested in further cooperation. He'll have to speak to
that after he's back, though.

> Chuck> It sounds like this works, to me. Things only get sticky if
> Chuck> Cygnus starts to do independent Apache development, and then if
> Chuck> I understand correctly, ServerName would have to change, or
> Chuck> Cygnus changes would have to fold back into the main Apache
> Chuck> development branch. So far, Gordon hasn't indicated that Cygnus
> Chuck> is interested in doing this:
> Let me offer my perspective on this, which of course does not
> represent actual policy, etc, etc: Cygnus probably will end up doing
> independent development, but we'll send our changes back to the Apache
> group.  If our changes meet with approval, then fine.  If not, but we
> still need them, then we will include them in our release.  In my
> opinion in this case it wouldn't be too hard for us to either make our
> patches available for ftp, or to put any modules we write into the
> module repository.  That way anyone could build an Apache like ours
> with a little effort.  (Whether or not we do this is not up to me, but
> I think it is a good thing to do)
I would think so, too. There's been expressed willingness from both
Apache folks and Cygnus to see this happen.

> I guess the question is: when does Apache not become Apache?  Suppose
> we incorporated some modules from the module repository into our
> release.  Would we still be able to call this "Apache"?  Or must we
> change the name in this case?
I think (and this is purely IMHO) that the server you suppose in your
examples above is still Apache, and should be called such. Non-core
modules would be at your, or anyone else's discretion to add, but you'd
be more on your own or dealing with the third party module developers
in terms of those. Randy's repository makes this rather less a sticky
proposition, though. The idea of a common store for other changes
via FTP is a good one.

> My understanding is that it isn't our intention to try to make a
> proprietary version of Apache.  Far from it -- Cygnus has a documented
> history of working with other free software developers.  However,
> sometimes our idea of what is necessary and useful will differ from
> other peoples' idea, and so there must occasionally be some source
> divergence.
That's my impression, too, and it's very reasonable. I can't claim to
speak with authority for the Apache group, since the elders are mostly
all in Paris right now. However,  I'd say this discussion, which started
out talking about version labelling semantics, should suffice to show
that Cygnus is on the level, and reasonable in its aims and its treatment
of others' works. That should be visible even for folks with no prior
knowledge of what you all do.

The naming conventions don't look insurmountable, and neither do the
codevelopment issues. I think it's important to realize how the two
groups compliment each other, and work from there.

This isn't Microsoft and Seattle Microsystems. 8^) Thanks for the
viewpoint, Tom. It's probably time to pause and let the rest of the
group ruminate on what's been said so far, and head onward from there.

Chuck Murcko	N2K Inc.	Wayne PA
And now, on a lighter note:
There exist tasks which cannot be done by more than 10 men or fewer
than 100.
		-- Steele's Law

View raw message