The following behaviour has always confused me.  It sounds like it has
"skipped" reverting the rm step, i.e. it will add lines to a file that
does not exist!

    $ rm x
    $ darcs rev x
    Reverting changes in "x"..

    hunk ./x 1
    -line1
    -line2
    -line3
    Shall I revert this change? (1/2)  [ynWsfvplxdaqjk], or ? for help: y
    Skipped revert of 1 patch.
    Do you really want to revert these changes? y
    Finished reverting.

Why doesn't it ask about the rmfile first?  That is,

    $ rm x
    $ darcs rev x
    Reverting changes in "x"...

    rmfile ./x
    Shall I revert this change? (1/2)  [ynWsfvplxdaqjk], or ? for help: y
    hunk ./x 1
    -line1
    -line2
    -line3
    Shall I revert this change? (2/2)  [ynWsfvplxdaqjk], or ? for help: y
    Do you really want to revert these changes? y
    Finished reverting.

The basic idea is that darcs revert asks for patches in the same "reverse" order
as darcs obliterate does (I think).

I'm going to mark this wont-fix for now, but Florent, weren't you also saying
something about this not being the best order?  I'm a bit hesitant to tinker
around with these things because what will likely happen is that some *other*
unintuitive-but-correct behaviour will emerge from it.  But I'll leave this up
to the rest of you.

I remember being puzzled by this order when i looked at reverse's code, but I
don't remember if there was a compelling reason for it or not.