The Translate task lets you pick a start token and an end token to delimit keys, but they must be single characters. This single character limitation makes it hard to pick a good delimiter character when you're running Translate against, say, an XSLT -- there just aren't many single characters left. Also, the Task actually has more internal conditions that are required before a string is considered a match. If I'm using # as my tokens, it corresponds to something like the regular expression: #[^\s=:#]# It might be nice to allow the user to specify a regular expression to determine what a string keys may look like, to reduce spurious matches.
The start token and end token can now be multiple characters long.