Also, there’s a macro that does something like define-syntax-rule, but uses syntax-parse’s functionality: define-syntax-parse-rule.

(require syntax/parse/define) (define-syntax-parse-rule (dada-test place-term ({~literal true-when} read-holds-atomic-term ...)

There’s also the option of having your dada-test macro use Racket keywords for these, like #:true-when and #:false-when. As a matter of taste you might love or hate how they look, or you might not care either way. But they’re unambiguously keywords.

PS for #:literals from syntax-parse the keywords need bindings for-syntax; if you don’t care about that, use #:datum-literals instead.

IIUC, those literals should not be required for-syntax. But they do need to be required normally and cannot be unbound. Their bindings are compared at phase 0 (which is the runtime one).

Ah, yes, perhaps it’s just define-syntax at phase–0—that seems closer to what I’m remembering

+1. It also works for normal define but using define-syntax can generate error messages for out-of-context uses.

I really like ~literal over #:literals, but there is a difference: ~literal allows the id to be unbound. When the id is unbound, the id used in macro invocation must also be unbound, but with the same name

So they are really merely compared by free-identifier=?!

apparently