William: Do you have a workaround for now, such as reordering the
definitions?
I have a change that makes these examples work, but it breaks other
examples, like a single expansion of this module:
(module namespace-mismatch racket/base
(#%plain-module-begin
(#%require (for-syntax racket/base))
(begin-for-syntax
(let ([ns (variable-reference->namespace (#%variable-reference))])
;; The top level at phase 1 ...
(eval #'(define-syntax-rule (m) (begin (define x 2) x)) ns)
;; The expander will have to find the right macro-introduced `x`:
(eval #'(m) ns))
(#%plain-lambda () foo))
(begin-for-syntax
(define-values (foo) #f))
(module* f #f
(#%plain-module-begin))))
The problem is that the current expander tries to do different things
in a module and at the top level to handle macro-introduced
identifiers. That seems ok for phase 0, where a module and top-level
process don't overlap for the same namespace, but these examples
illustrate how it fails for phase 1.
The change to make other examples work is at
https://github.com/mflatt/racket/tree/phase-1-ns
Since it may break existing programs, though, I'm reluctant to push it.
The right solution is to switch to the new expander implementation ---
soon, I hope. The new expander works on all of these examples, and it
generally gets `...->namespace` operations right where the current
expander's implementation is flaky.
Thanks all for very helpfully narrowing down the problem!
--
You received this message because you are subscribed to the Google Groups
"Racket Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to racket-dev+unsubscr...@googlegroups.com.
To post to this group, send email to racket-dev@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/racket-dev/5a75d66b.494f370a.ff7f0.37d7SMTPIN_ADDED_MISSING%40gmr-mx.google.com.
For more options, visit https://groups.google.com/d/optout.
