Unless I supply a --limit, darcs changes -i is very slow:

    $ time darcs cha -i GNUmakefile --last=100 <<<q
    Mon Apr  6 19:29:52 EST 2009  Trent W. Buck <trentbuck@gmail.com>
      * Remove autoconf references to patch_theory.pdf, which is subsumed by haddock.

    Shall I view this patch? (1/4) [yNvpxqkj], or ? for help:

    real    0m0.245s
    user    0m0.040s
    sys     0m0.090s

    $ time darcs cha -i GNUmakefile <<<q
    Mon Apr  6 19:29:52 EST 2009  Trent W. Buck <trentbuck@gmail.com>
      * Remove autoconf references to patch_theory.pdf, which is subsumed by haddock.

    Shall I view this patch? (1/592) [yNvpxqkj], or ? for help:

    real    0m24.785s
    user    0m2.800s
    sys     0m2.520s
    $

I believe that Darcs should be lazy about this, i.e. it should talk to
me about the first matching patch straight away, without waiting to
tell me about the other patches.

I guess the actual problem is that Darcs wants to work out the number
"592" above, before printing the first prompt.  I remember that for
similar cases darcs does/did print (1 of ?), and there was a separate
prompt command "c" (count) which would eagerly evaluate and change the
"?" into a number "592".

The following patch updated the status of issue1430 to be resolved:

* Resolve issue1430: lazy pattern matching in (-:-) from Changes command module. 
Ignore-this: d1d3a01a9480cd142bd1f06ec24d6ed6

Prevents darcs from computing full patch list for interactive changes.