Re: Arch as a replacement for CVS for OpenBSD?

[Tom Lord <lord@regexps.com>]

       A B <obsdfaq30@yahoo.com> writes:

       > Is anyone considering arch, http://www.regexps.com/#arch,
       > a replacement for CVS as the OpenBSD repository tree?

I wanted to correct some errors and misleading statements:

	Artur Grabowski writes of arch:


	The feature comparison is made by the author.

And, in fact, contains corrections requested by the Subversion
developers.  Yes, it is intended to focus on the strengths of arch --
not to be a broad survey of all systems available.


	The documentation of the project is very confused and mixes slandering
	of other projects with ramblings about open source philosophy and some
	description about how to use it.

I hope other people won't take your word for that, but will check it
out for themselves.  Several users have found the documentation to be
very effective, others find it confusing, personally, I think it's
somewhere in between.  I don't think there's anything in it that fits 
the inflammatory parts of your description.


	It's not really clear to me how the repository works, but the
	more I look at it the more I suspect that everyone must have
	an own copy of the repository, making this a glorified variant
	of rcs.

You are completely wrong which is pretty much what should be expected
if you are trying to dismiss a project that you yourself admit you
don't understand and are unwilling to learn about.

Arch is a distributed revision control system, meaning that projects
(branches etc.) can span repository boundaries.  Any developer or
group _can_ have their own repository, which is very convenient in
many circumstances.  No developer _must_ have their own repository.


	This is not a replacement for cvs. A replacement would have a
	small transitional cost. arch looks completly different from
	all other revision handling systems I've seen. It's so
	different that I won't even waste more time trying to
	understand it.

It is very different from CVS.  It takes a very different approach.
Consequently it is smaller and simpler, yet has very useful features
not found in CVS (such as atomic commits, support for renaming files,
high-level operations to ease the difficulty of merging branches, and
distributed repositories).


	    But don't even imagine that OpenBSD will switch the
	    revision handling system within a year or two. Something
	    like that must be done very carefully to a system that's
	    tested. If we'd want to do it at all.

At least we agree on something (the need for care and caution).

-t