










@samth I think we’re writing into “compiled” dir instead of “pycket-compiled”

you need to update racket


I did, I’m using 1.7.0.2

7.1.0.2 that is

do you have aa6ec1f0cd51ce630c57c2bf343eb6f911077ffc ?

Yes, I do have that commit from 3 days ago

(have to leave now but I’ll look into it)

Ah I think it’s because we’re not setting the use-compiled-file-paths
in compile-file-pycket.rkt
module

Shouldn’t it happen automatically when we run under pycket?

But we’re resetting the namespace in compile-file-pycket

│··················
[samth@huor:~/sw/plt (master) plt] pycket-c-linklets │··················
Welcome to Pycket v7.1.0.1 │··················
> (use-compiled-file-paths) │··················
(#<path:pycket-compiled>) │··················
> (current-namespace (make-base-namespace)) │··················
> (use-compiled-file-paths) │··················
(#<path:pycket-compiled>) │··················
>

@dan has left the channel

@samthYes, it’s not about the namespace, it’s about that parameter being overloaded to contain two pieces of information. The recompile
target in our Makefile clears the zo files and build them again. And we run the clean
with -c
flag (to avoid getting version errors when we bump up the expander version), and -c
sets the parameter to null to basically say “don’t use compiled code”, so the rebuilding uses the new value null
and compile-file
defaults to writing into "compiled"

Ah

We basically need to reset the parameters value after clean

So we need to use parameterize more

Or really just make the regular Racket tools work

Did your change fix the unbound variable errors?

It seems so, but due to this confusion I’m not entirely sure

I’ll fix this parameter issue and run everything again
