Flexible File Synchronization?

lamour asks: "I have a Solaris desktop at work, a Linux desktop at home, and a Linux laptop. There are some files I want synchronized on all of them (think dot files), some that I just want synchronized to the laptop (think taking work home with you), and obviously lots that I don't want synchronized at all. Depending on what I'm working on, the files that I want synchronized may change (for security reasons, under no circumstances do I want any files on my laptop that don't need to be there, so ease of reconfiguration is key). I'd be willing to concede that the "changing" portions of the synchronization could probably be managed at a directory level. It occurs to me that the n-way, full synchronization, and the two-way, temporary synchronization could and maybe should be maintained by separate tools. I have some thoughts, but wanted some other input on this issue. Anyone know of any tools or hybrid of tools which would make this painless?"

coda

[jab]

Try the Coda distributed filesystem, and its "hoarding" mechanism. If you are brave.

Strange Answer: CVS

[Falsch Freiheit]

I've found myself with some somewhat similar needs. The solution that I find works best is a private CVS repository on the most accessible machine in the bunch.

You can set up CVS with "modules" that will cover a very specific list of files that you can change by editing the modules file in the CVSROOT (just cvs checkout CVSROOT, edit the file in there and commit).

You can then have some symlinks into the directory checked out of CVS or do a little trickery and get your home directory to look to cvs like you checked it out.

With text files, you can then possibly make changes on two different machines and merge the changes together. With non-text files you need to be careful to commit changes and update everywhere relevant before changing the file someplace else.
(Make sure to mark binary files as binary files, BTW)

Also, it's pretty easy to get cvs to work via ssh