[Tim]
That would be true if the standard explicitly forbade extensions (any optional operation or behavior not specified by the standard). But there are no standards like that ;-) To comply with a standard, there just needs to be _some_ way to "spell" all the behaviors the standard mandates.
[Mark H. Harris]
The IEEE 854 1987 is just such a standard (Mike Cowlishaw's extension).
The generalized IEEE 754 2008 standard might not have as much wiggly room.
I have the 2008 version of the standard; it's thoroughly ordinary in these respects. As usual, it's not even "a language" that's normally said to be compliant with the standard. Instead: Conformance to this standard is a property of a specific implementation of a specific programming environment, rather than of a language specification. Although: However a language standard could also be said to conform to this standard if it were constructed so that every conforming implementation of that language also conformed automatically to this standard. So far as languages go, there's very little a standard of this nature _can_ say! Languages vary enormously in syntax and conventions. Indeed, for example, the standard "spells" multiplication as multiplication(x, y) That doesn't mean a conforming implementation has to supply a 2-argument function named "multiplication". It just means a conforming implementation has to supply _some_ way to spell what the standard defines via "multiplication(x, y)". Etc.