Part 4 Annex A.3 formally describes the syntax of NumericRange as follows:

<numeric-range> ::= <dimension> [',' <dimension>]
<dimension> ::= <index> [':' <index>]
<index> ::= <digit> [<digit>]
<digit> ::= '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9'

This would allow for a maximum of two-dimensional arrays. Also, any index could have only two digits, and leading zeroes in an index would be allowed (I am not sure whether the latter is intentional or not).

As I believe these limitations are not intentional, I am proposing the following set of rules as a possible solution addressing the problems:

<numeric-range> ::= <dimension> [’,’ <numeric-range>]
<dimension> ::= <index> [’:’<index>]
<index> ::== ‘0’ | <digit> [‘0’] [<index>]
<digit> ::== ‘1’ | ‘2’ | ‘3’ | ‘4’ | ‘5’ | ‘6’ | ‘7’ | ‘8’ | ‘9’

Note: the notation used is not BNF, so we should probably either use BNF, or call it something else.