1.2.0

Enhancements:

• A new search operator, :, now enables capturing slices of Arrays (by 0-based
  element number) and Hashes (by alphanumeric key-name). This looks like:
  "some::array[2:15]" or "some::hash[beta:gamma]".
• yaml-get now returns JSON instead of "pretty Python" data objects when the
  search returns complex data types (Arrays and Hashes). This change makes the
  result more portable to non-Python consumers and ensures the result will be
  one per line.
• The separator used for identifying Hash sub-keys can now be customized. If
  you prefer your paths to look like "/hash/sub/key" rather than "hash.sub.key",
  you can now have it your way. For now, only . and / are allowed. The
  seperator can be either strictly specified or automatically inferred by
  whether the first character of a given YAML Path is /. Command-line tools
  like yaml-get and yaml-set have a new --pathsep argument for this; the default
  is "auto" and can be set to "fslash" (/) or "dot" (.).

Bug Fixes:

• EYAML on Windows now works when a batch file is used to wrap the Ruby eyaml
  command.

Known Issues:

• Escape symbols in YAML Paths parse correctly and will be properly processed,
  resulting in retriving or setting the expected data. However, the parsed
  path cannot be stringified back to its original form (with escape symbols).
  This issue affects only logging/printing of the post-parsed path. A unit test
  has been created to track this issue, but it is marked xfail until such time
  as someone is willing to tackle this (very) low priority issue. Until then,
  developers should try to print the pre-parsed version of their paths rather
  than rely exclusively on Parser.str_path(). Further, don't do this:
  1. Accept or take a string path that has escaped characters.
  2. Parse that path.
  3. Stringify the parsed path.
  4. Parse the stringified, parsed path. This is silly, anyway because you
     already have the first (good) parsed result at step 2.
  5. Try to use this parsed-stringified-parsed path result for anything.
     Instead, only use the first parsed result that you got at step 2.