A bunch of clean up/refactorings of Commands/Replace and Patch/TokenReplace,
followed by a fix for issue2125.


7 patches for repository http://darcs.net:

Thu Mar 29 22:06:40 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * Tidy Commands/Replace

Thu Mar 29 22:10:17 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * camelcase a few identifiers in Commands/Replace

Thu Mar 29 23:52:59 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * Tidy Patch/TokenReplace

Thu Mar 29 23:54:54 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * Extract common token-breaking code in Patch/TokenReplace

Fri Mar 30 00:11:59 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * Refactor/tidy-up Commands/Replace

Fri Mar 30 00:33:06 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * Accept issue 2125: darcs replace always warns about forcing, even when unnecessary

Fri Mar 30 01:17:48 BST 2012  Owen Stephens <darcs@owenstephens.co.uk>
  * Resolve issue2125: only warn about forcing replaces, when a force is required.

Attachments

• patch-preview.txt
• tidy-commands_replace.dpatch
• unnamed

Yup, looks good (I didn't understand the gap stuff, but apparently that's 
not a big deal).

Three silly style details I think we could follow up on (that I'm feeling 
a little less shy now we're cleaning up)

1. Before

     foo <- long long thing x
     bar <- long long thing y

   After

      [foo, bar] <- mapM (long long thing) [x,y]

   Better yet?

       let shorter = long long thing
       foo <- shorter x
       bar <- shorter y

2. Keep scope small (the less knowledge the better)

   Before: do
      blah blah
      let x = does not actually depend on blah blah
      other blah
 
    After: do
       blah blah blah
       other blah
     where
       x = does not actually depend on blah blah (less chance of 
error/misunderstanding?)

3. Possible chance to use the case () of _ | pattern as a better refactor 
for the nested if

Looks like this bundle had already been pushed.