Synchronisation -- when files can change on both sides

All the tools we saw till now were either mirroring, or backup, solutions. It was very clear that one side of the backup was the "source", and that data always moves only one way unless you're restoring from a backup because you lost something.

Unison, the last tool we will look at, is for a slightly different purpose -- synchronising two directories that may have changes independently made on them. It is also the only one with a GUI, and a pretty good one at that! And a GUI is needed when synchronising, because there may be conflicts that need to be resolved.

When is it useful?

Since I don't have a laptop, and tend to work parallelly on my home machine and work machine during the week, every weekend I just sync the "work" directory on my home machine with my USB flash disk, and then I sync the "home" directory on my work machine with the same USB flash disk.
Yes -- you can synchronise A to B, then B to C, and back again at regular intervals, and it will have the effect of keeping A and C synchronised. Pretty cool! In this case A is a directory on my work machine, B is my USB flash disk, and C is a directory on my home machine.
Without Unison, the only way to do this easily is to work directly off the USB disk. More convoluted schemes are possible, but could be hazardous to your sanity :-)
For ad hoc use, it also makes a very efficient, interactive way to compare two directories and synchronise them. The GUI really helps here!

How is it used?

I won't describe the batch mode at all, which should tell you how good the GUI is :-) Also, synchronisation ideally should not be batched anyway, since manual intervention may be needed.

However, Unison does have a "profile" concept which you must use in order to get the best out of the tool. The profile is a plain text file, with a fairly simple syntax, and is not at all difficult to understand.

Unison calls the pair of directories you're trying to synchronise the synchronisation roots, and sub-directories/files within the roots are called paths. Most of the profile is about including and excluding paths.

[Come to think of it, this whole include/exclude business is what takes the maximum time to understand in any tool like this!]

Starting Unison

Just type in the command. The first time, it will ask you for the names of two directories to compare, and create a default profile for these two in your $HOME/.unison/default.prf. If you have a .unison directory with any .prf files in it, starting unison will give you a screen listing all your profiles. Pick one, and start synchronising.

Make a couple of test directories in /tmp or something and practice a little till you get used to the GUI, and the keyboard shortcuts, which are all conveniently given next to their entries in the GUI's menu system. It's really very, very, simple! The online help is also pretty good -- all the documentation is available from within the program.

You can create profiles manually using any editor. I recommend this -- you can control things much better. Let's look at a few profiles now.

Sample profile 1

This is a simple one meant to transfer files between a directory on my hard disk and a USB disk:

    label = sync the a780 flash disk

    include default

    perms = 0o0400
    pretendwin = true

    root = /home/sitaram/a780-backup/
    root = /mnt/removable

When you start the GUI without any arguments, the label strings of all the profiles unison found in ~/.unison are listed as choices. So put in some descriptive text here.

The include is easy to understand. By the way, my default.prf is:

    # Unison preferences file
    height = 40
    diff = gvim -d

This just says that I prefer to use gvim -d to be invoked when I ask the GUI to show me the differences between two files (for example if they have conflicting changes).

The perms and pretendwin are needed because the flash disk is a Windows file system (VFAT) which does not support many Unix file system semantics. Read the documentation if you want to know more.

The last two lines just say what directories I am synchronising with this profile.

Sample profile 2

Here's a somewhat more complex one. Note that, unlike rdiff-backup (but like dar), the order of ignore (exclude) and ignorenot (include) does not matter, which I like very much.

    label = RC files and other settings

    include default

    root = /home/sitaram
    root = ssh://sitaram@remote.host

    # "ignore = Path something/*" works, but "ignore = Path *" doesn't
    # (this is documented!)
    ignore = Path ?*
    ignore = Path .*

    ignorenot = Path .*bashrc
    ignorenot = Path .darrc
    ignorenot = Path .vimrc
    ignorenot = Path .sunrise
    ignore    = Path .sunrise/cache

What are the downsides?

Conflicts and reconciliation -- although for text files you can do a pretty decent job of merging conflicts if needed. Just use gvim -d!
If you are doing this across an SSH connection, the other side should also have the same version of Unison installed.

Additional notes

You can ask unison to show you a diff between two files. See above for my preferred setting for the diff tool.
You can interactively add to its preferences -- explore the GUI to see how, then open the profile file in an editor to see the lines that were added by the GUI.
You can also ignore files with certain names, regardless of where the name occurs. Example:

ignore = Name {CVS,*.cmo}

Oh, and did I mention that Unison also uses the rsync algorithm internally to transfer files to the other side when used across a network?

:-)

What next?

Well, we're really done, actually. However, as a bonus if you made it this far, here's a summary of most of the stuff we've seen so far, which I keep as a quick reference: Summary of tools, syntax, and notes