> So what *is* the goal of SRFI 220?

- Using S-expressions to represent the set of magic directives that
parse easily as S-expressions.

- Not doing anything about the other magic directives, as once you go
down that road, there's no end to how hairy and ad-hoc they become.
Hopefully the job of those directives can be done in an alternative way.

- Ideally, unifying other people's magic directives with RnRS directives
into a common framework.

- If Scheme ever gets Clojure-style metadata
(https://clojure.org/reference/metadata), it would be great if that
could be rolled into the same framework as well instead of having a
different syntax for it.

> If you don't want to allow arbitrary
> datum formats, then we don't need the line ending delimiter that doesn't
> fit Scheme datum syntax and the implicit wrapping in lists.
>
> Just declare shebang lines special (they can only be present at the
> beginning of a file) and otherwise specify a syntax of the form:
>
> #! <datum>

It may be useful to have them elsewhere in the file, e.g. at the end,
not only at the beginning. And reading some syntax only at the beginning
of a stream is a similar irregularity as parsing until end of line; I
think the latter is a bit simpler.

The #! <datum> format is probably something we should have eventually in
any case, but how exactly to combine it with the present proposal is
still unclear.

> Whether this is good coding or not, it is a possible shebang line that
> would make SRFI 220 break.

If you're talking about the kinds of hacks where the same script doubles
as a shell script and a scheme script, there has to be a better way.

>     The portable Unix shebang lines are basically:
>
>     #! /bin/sh
>     #! /usr/bin/env foobar
>
> Change my example to
>
> #! /usr/bin/env (-scheme

What does the example mean? Unix commands should have alphanumeric names
(optionally with dashes, dots, and underscores). There are lots of
things that don't cope well with other stuff.