Implementation details
compact-string

==The encoding==
Data.CompactString encodes characters using a variable length encoding,
a single character is stored as 1, 2 or 3 bytes.
The bytes are stored in a ByteString, so:
> newtype CompactString = CS { unCS :: ByteString }

The encoding used is not UTF-8, because that is rather inefficient.
Characters in Unicode contain at most 20 bytes, these are encoded as:
  <table>
   <tr>
    <th>Range</th>
    <th colspan="3" class="x">Character</th>
    <th colspan="3" class="x">Encoded</th>
   </tr>
   <tr><td>U+0000 - U+007F</td>
       <td class="x"><br></td><td><br></td><td>0zzzzzzz</td>
       <td class="x"><br></td><td><br></td><td>0zzzzzzz</td></tr>
   <tr><td>U+0080 - U+4000</td>
       <td class="x"><br></td><td>00yyyyyy</td><td>yzzzzzzz</td>
       <td class="x"><br></td><td>1xxxxxxx</td><td>0zzzzzzz</td></tr>
   <tr><td>U+4000 - U+10FFFF</td>
       <td class="x">000xxxxx</td><td>xxyyyyyy</td><td>yzzzzzzz</td>
       <td class="x">1xxxxxxx</td><td>0yyyyyyy</td><td>1zzzzzzz</td></tr>
  </table>
The tagging bits might seem strange at first.
They are chosen so that the character can be read both front-to-back and back-to-front.

See <a href="/repos/compact-string/Data/CompactString/Base.hs">Data.CompactString.Base</a> for details of the encoding.

==More to come?==
''todo''