[relaxng-user] Documentation tokenization

[Jeff Rafter]

Sorry for the endless stream of questions. I wanted to do a quick sanity
check on documentation tokenization. I have been testing my theories against
XMLDistilled (which is extremely useful) and they are not lining up.
Roughly, this is the step by step for reading documentation:

A. Require "##"
B. Skip additional "#"
C. Single space is required before additional content (e.g. anything not
NEWLINE or not #xA;)
D. Begin reading content (until NEWLINE_MARKER or #xA;)
  D.1 Cannot contain "#"
E. Is the line continued?
  E.1 Read past one NEWLINE_MARKER or #xA;
  E.2 Skip Indent (#x9;#x20)
  E.3 If "##" goto B

The main questions I have are:

(1) Is step "D.1" accurate? from documentationLineContent production in A.2
it seems so. It appears that XMLDistilled allows additional "#" after
content

(2) Is step "C" accurate? from documentationLineContent production in A.2 it
seems so. Again, XMLDistilled allows this.

(3) Is step "E.1" correct? I presume it makes little difference in the large
scope:

  ## some annoation

  ## more annotation

Ultimately this would be combined to a single documentation. My curiousity
on this point is simply about correctness.

Thanks,
Jeff Rafter