[OT] Re: CVS Access Lists WAS Re: [PD-dev] sourceforge tarball problem

Hans-Christoph Steiner hans at eds.org
Thu Nov 7 22:39:01 CET 2002


On Thu, 7 Nov 2002, Chris McCormick wrote:

> Infact, I'd say that the whole basis of open source development is about social protocols rather than programatic protocols. Anyone
> could take the sources to the Linux kernel or Pd and rename it to "McCormix" or "unpure-data" a lot like what happened with Xemacs.
> But as long as the project has good maintainership like Miller or Linus provide, people will continue to follow correct social
> protocol and defer their additions/modifications of the source to that central maintainer.
> In a way this is the perfect blend of democracy and socialism. If only we could design the code of governance and law in the same way
> maybe we'd live in a utopia. That or total anarchy, which might be just as fun.
> [/perhaps]

Sure, but on the large, successful projects, those social protocols are
enforced by maintainers.  Look at how Linux, Debian, GNOME, KDE, GNUstep,
etc. etc. ad nasuem work.  They are all broken up into sections with
maintainers controling each section and permissions to enforce that.  

If you don't have permissions to edit a file, then you submit a
diff to the maintainter.  If you start doing a lot of work on a section,
then the maintainer gives you access.  Alan Cox is the perfect example of
this.  But I think that even Alan Cox doesn't have full access to the
linux sources (I could be wrong).

I am not advocating this because I want to see pure-data carved up into
little fiefdoms.  This is a tried and true structure for developing
software with many developers contributing.  Plus it could help MSP to get
the pd code into CVS sooner, since he could work off of CVS alone, without
having to develop the test suite needed to allow multiple developers.

.hc

	zen
	   \
	    \
	     \





More information about the Pd-dev mailing list