What is the formal definition of a SQL identifier in the ISO/IEC 9075:2016 standard?
Most specifically, in the case of both regular and delimited identifiers, what limitations exist on:
You can look up the rules e.g. here
So a "regular" (i.e. "non-delimited") identifier must start with a <simple Latin letter> which is essentially only the characters a-z (either upper or lower case).
Then it maybe followed by an <SQL language identifier part> which is either a <simple Latin letter> or a <digit> or the <underscore>
To sum it up: a regular identifier must start with a letter from a-z, and can only contain letters, digits (0-9) and the underscore.
A regular expression to validate a simple identifier would be: [a-zA-Z][a-zA-Z0-9_]*
Mandating "simple latin letters" makes any multi-byte character set (e.g. UTF-8) invalid for use with a simple identifier.
Other character (e.g. a space) can only be used when using a "delimited" (aka "quoted") identifier which is enclosed in double quotes, e.g. "%foo bar$". As far as I know this is also true for multi-byte characters.
The allowed length of an identifier is discussed in this question: Does the SQL-1992 standard restrict naming identifiers to 18 characters?
