Hello Racketeers,
while working on new Actor Basic version I stumbled upon a strange behavior. I managed to narrow it down strictly to my rrll-abasic package (not in the package catalog) sources found at rrll-abasic · master · Dominik Joe Pantůček / rrll · GitLab
A minimal non-working example with the current version is:
#lang s-exp rrll/abasic/standalone
(define (test-actor))
The result is:
mwe.rkt:3:9: test-actor: unbound identifier
in: test-actor
location...:
mwe.rkt:3:9
Here, define is just rename-out of define/provide-actor defined in rrll-abasic/abasic/expander.rkt which is pretty innocuous and was like this since long long time ago.
The error goes away if I remove space.rkt from standalone.rkt (line 4) and the (run-steps run) rename-out (line 22) - as it comes from space.rkt.
And yes, I have been working on space.rktlately to provide an external access to the evaluation environment from other threads - I added the last three procedures in that file and updated the thread field in run-actor (line 177). Just to give a complete picture, I added the fields thread and cmd to the %space struct in space-struct.rkt as well.
Perhaps I am overlooking some simple explanation, by my overall impression is that for some reason the current space.rkt "confuses" the Racket expander and I get strange behavior in other code that transitively requires it.
As a side note - it originally expressed by not evaluating any code so I tried adding some runtime debugging logging and errortrace - but adding errortrace to the expander.rkt module resulted in:
namespace mismatch: bulk bindings not found in registry for module: #<resolved-module-path:"/home/joe/Projects/Programming/RRLL-NEW/msg.rkt">
context...:
/home/joe/racket/collects/syntax/wrap-modbeg.rkt:46:4
I will file a bug report if nothing obvious shows up, of course.
Cheers,
Dominik