> Comments are not scheme datums. That's the point of comments.

The point of comments is that they are the one language construct
dedicated to the purpose of not being machine-interpreted for meaning.

As Marc points out, Scheme has #; datum comments, so a comment can be a
datum. This feature is somewhat idiosyncratic to Lisp and Scheme, but
it's in line with the logic that comments are discarded by the computer.

Magic comments are an inherently bad idea because they are comments that
are _not_ discarded by the computer. When programmers build on such an
illogical foundation, it's no wonder that the stuff built on top of it
ends up being a pile of hacks as well.

Encouraging generality on top of this foundation just makes this
ill-advised pile bigger. What we should encourage is that people move
functionality out of this pile into something based on solid engineering
principles, such as not mixing two syntaxes in one file.

Marc is pointing out we are now mixing two syntaxes (a line-oriented one
and a nested, whitespace-oblivious one), and that's a good warning. But
he suggests another approach that mixes syntaxes in an even worse way :)

> If they could be written in scheme datums, they would have been Scheme datums.
> But they are not. Or, rather, they may be Scheme strings, as far as I understand.

Line comments (;...) are like Scheme strings.

Multi-line comments (#|...|#) are like multi-line Scheme strings.

Datum comments (#;...) are like Scheme datums.

Most languages don't have datum comments, so any confusion is
understandable.

> Aren't directives actually comments?

The point of directives is to have an effect, but an effect that is
different in some way than evaluating the datums that follow the #! as
ordinary Scheme code. Marc gave good examples.

The point of comments is to be ignored by the computer.