Thanks, applied.

Stone

Looks good to me.  I wonder if in the future it might be necessary to
return a bool from fs.mkdir indicating success.  Thats not necessary now
though.  Applied.

Stone

Thanks, applied

Stone

LGTM, thanks.

Stone

>  tests/project/meson.build             | 1 +
>  tests/project/muon/python/meson.build | 9 +++++++++
>  2 files changed, 10 insertions(+)
>  create mode 100644 tests/project/muon/python/meson.build
>
> diff --git a/tests/project/meson.build b/tests/project/meson.build
> index dc973e4c..593aed0f 100644
> --- a/tests/project/meson.build
> +++ b/tests/project/meson.build
> @@ -5,6 +5,7 @@ tests = [
>      # project tests created for muon
>      ['muon/timeout', ['failing']],
>      ['muon/sizeof_invalid'],
> +    ['muon/python'],

Thanks for this patch, I'm glad you were able to make sense of the weird
rcvr_transform stuff.  Overall it looks good, but I have a few comments.
I do have to ask, were you able to run the tests?  With this patch, for
instance, muon asserts when trying to setup itself.

> diff --git a/include/lang/object.h b/include/lang/object.h
> index 076c6cf7..5da15ce0 100644
> --- a/include/lang/object.h
> +++ b/include/lang/object.h
> @@ -39,6 +39,7 @@ enum obj_type {
>  	obj_subproject,
>  	obj_dependency,
>  	obj_external_program,
> +	obj_python_installation,

LGTM

Stone

LGTM

Stone

Thanks, applied.

Stone

Thanks for this patch.  Overall, I think with a few small changes we can
avoid modifying const arguments.  This is something that I always try to
avoid when possible as it makes the code more difficult to understand.
If it is truly impossible to avoid, then all of the function signatures
should remove const from the modified arguments to correctly indicate
what is going on.  Exceptions may be made for calling into external
libraries e.g. execv, but inside code we control, constness should be
kept as consistent as possible.

Stone

> @@ -69,6 +69,7 @@ _path_normalize(struct workspace *wk, struct sbuf *buf, bool optimize)
>  		return;
>  	}