Zefram wrote:
Ed Murphy wrote:
I'm thinking "SHALL, unless e reasonably believes that assigning a
smaller number might be invalid".
Too tight. If a number assignment has been incorporated into persistent
documents, such as a published ruleset, I shouldn't have to reuse it if
the entity numbered turns out not to have existed.
In most such cases, some higher numbers would have already been
assigned. Still, may as well change "might be invalid" to "might
be invalid or confusing".
I have an idea for preventing the use of really colossal numbers: require
that the ID number being assigned be stated explicitly as a decimal
literal in the assigning announcement. No chained arrow notation for us.
On top of that, we could cap the number of digits based on the
number of digits in any previous number (e.g. cannot expand to
5 digits until at least one number 9xxx has been assigned).
