#629 dollar sign escaping in canonical references

[Hugh A. Cayless]

At the end of http://www.tei-c.org/release/doc/tei-p5-doc/en/html/SA.html#SACR it states: "If there is a need for an actual string including a dollar sign followed by a digit that is not supposed to be replaced, the dollar sign should be written as %24." This is problematic because '$' is a reserved character according to RFC 3986. It's a so-called "sub-delimiter" (though not a commonly used one). What this means is that '$' in a URI will not necessarily be treated the same as '%24', and so we shouldn't recommend that substitution.

It would be better either to escape the '$' the way most regex processors do, as '\$' or '$$', or to use '\1', '\2' etc. to reference capturing groups and recommend escaping that as '%5C'.

[James Cummings]

Assigning to Hugh Cayless to bring to Council for quick explanation, vote, and implementation.

[Hugh A. Cayless]

Had to re-read this to remind myself of what it's about. The upshot is that '$' is a special character in URI syntax, and is not normally escaped. '%24' in a URI would mean 'a dollar sign not being used as a sub-delimiter'. Therefore '%24' and '$' might not be interpreted identically by software that has to parse URIs. The simple solution is to pick a different escape syntax for '$' and recommend that any Canonical Reference interpreter substitute that at the same time as it handles '$n' back-references. I like '$$'.

[Syd Bauman]

Council agreed to '$$'

[Hugh A. Cayless]

Fixed with r12919.