06:27 <taw10> nuance: I'm not sure, but I suspect it might be related to this issue described in the Guile manual: https://www.gnu.org/software/guile/manual/html_node/Hygiene-and-the-Top_002dLevel.html

06:28 <taw10> nuance: Basically, you can't create a situation where the module ABI changes programmatically, but you can get away with it if the compiler can see that the bindings you're making aren't going to change. However Racket probably handles this situation in a different way to Guile.

11:24 <nuance> taw10: Thanks. My gut instinct is that you're more or less right--that this has to do with hygiene and/or phase level. I've tried variations on local-require and regular require, and there *are* some pointers on Stack Overflow (e.g. https://stackoverflow.com/questions/18852921/generating-require-clauses-with-racket-macros), but they aren't exactly applicable. The use of body ... in the macro is

11:24 <nuance> what's likely making life miserable at this point.

11:24 <nuance> I'll keep poking at it, but would welcome any additional thoughts if someone in the channel has them.

13:49 <samth> nuance: the problem is hygiene; if you change `make-module` to be `(make-module nm lang body ...)` then it will work

13:49 <samth> alternatively you can use `datum->syntax` to introduce the lang position with the correct scopes

16:06 <nuance> samth: ahhh, thanks. Glad it's a straightforward fix, but I'm definitely smacking my forehead on this one. I suppose on the plus side I managed to read a lot of documentation.

16:08 <samth> nuance: `require`, including the implicit one in the language position, is unhygenic binding and thus annoying to abstract over, so it's not too surprising to get tripped up

16:09 <nuance> samth: maybe it's a silly question, but is there any way to modify the language to make this hygienic (e.g., by default namespacing)?

16:13 <samth> If you want `#lang foo` to bind `lambda` inside the body (which basically you do want) then the non-hygiene is fundamental

16:14 <samth> you could make it be `foo.lambda` but then you've just made `lambda` a string instead of a reference and thus you haven't really added hygiene

16:14 <nuance> Nod. I'll have to noodle on this a bit. I understand basic hygiene (cue bad jokes about bad breath), but obviously there are some things missing for me. It smelled like a hyginee issue, but I couldn't quite make the jump to the binding there.

16:16 <nuance> i.e., what's happening in my understanding is that within the context of the transformer, `racket` means nothing, more or less?

16:16 <samth> you could imagine a language (eg a slightly modified java) where (a) keywords like `class` aren't "bound", but handled by the parser and (b) there's only import of specified names

16:17 <samth> then there wouldn't be this issue, but that would be very different from how Racket works

16:17 <samth> in the context of the transformer, `racket` doesn't have the same scopes as the body forms (in particular it has an extra scope from the expansion step)

16:20 <nuance> samth: that makes sense, thanks.

22:02 <runrin> Why is it that if I don't explicitly put ``#lang racket`` at the top of a file I load into the command line REPL, I can define procedures that reference each other out of order, but if I make the language explicit I can't?

22:03 <runrin> For instance, if I have ``(define (test x) (proc x x))``, if the language is set explicitly, ``proc'' needs to be defined before test.

22:04 <runrin> if it's not explicitly set, the order the procedures are defined makes no difference