PROPOSAL - Template Support #84

For those that have been around a long time, you know this discussion started in 2013 and never ended - primarily due to my inability to become comfortable with any of the proposal... especially my own :)

I've been mulling this over on and off for the last 2 years more than usual and it was always the intersection between simplicity and consistency with the existing spec that didn't quite click for me.

With that said, I would like to make the following proposal and see how it tastest/feels/smells to the rest of the community.

• [_][integer length][0 or more data-type markers] (e.g. [i][S][C])
• no container markers allowed in the template definition
• can occur 0 or more times inside of a container
• additional complexity in the parsers would be to check the 1st byte value after all expected structured data was read to see if it was 95 (another _ marker)
• In the case of [S] or [H], only the type marker is specified and the length will be pefixed down in the data payload; similar to how Object containers are defined to omit the [S] off the name portion of the entries.

Example:

[{]
  [_][i][3][S][C][i]
  [i][5][frank][Y][3]
  [i][4][mary][P][9]
  [i][7][dillion][J][13]
  [i][7][willard][W][22]
  [i][3][bob][V][11]
  [_][i][2][i][d]
  [7][3.14]
  [13][2.18]
[}]