using cvs without a server to distribute code

[Ed Sutter]

Hi,
I'm new to this list, so I start by apologizing if my question
is stupid.  I did a non-robust search through the archive looking
for an answer, and came up blank, so here goes...

I have a source tree for some embedded stuff that I provide to
folks who ask for it.  I had been using SourceSafe to keep track of
changes and I simply provided folks with a tar-ball.  They had no
record of the change history except to run a diff between their
current tree and a new one that I distribute.  Not good.

I've transitioned over to CVS for tracking changes, so I'm hoping that
I can now allow folks to track changes between the version of the source
they currently have and a new version that I distribute.  The problem
that I have is that I will not have a CVS server for folks to interface
to.  So, is there a common and/or recommended way to distribute code
that is under CVS without using the cvs server?  For example, I know I can
just tar up the repository and send that off to folks; however, does this
support their ability to be able to compare their code to the newly
distributed repository?

I'm hoping that users can apply the following steps:

1. Retrieve the tarball repository that I distribute.
2. Check out the latest version of the code under that repository.
3. At some later date they can retrieve a new repository that I distribute
   and replace their current repository with this new one.
4. Run "cvs diff" on their checked out tree (sandbox) vs the new repository
   to see what's changed.

Does this make sense?
Thanks for any help!
Ed