sagemath_environment must respect --enable-system-site-packages

[dimpase]

Steps To Reproduce

Trigger building sagemath_environment, e.g. (after applying #38163) by make sagemath_tdlib, or from scratch, by

./bootstrap && ./configure --enable-system-site-packages --enable-sagemath_tdlib && make build

then, assuming that either setuptools or wheel packages come from the system, it will fail, as observed on #38159 (comment)

Expected Behavior

should work

Actual Behavior

in does not.

Additional Information

There is a lookup happening for wheels for setuptools and wheel - something that is almost sure to fail if the latter come from the system.

Environment

- **OS**: Gentoo Linux
- **10.4.beta9+#38163**:

Checklist

• I have searched the existing issues for a bug report that matches the one I want to file, without success.
• I have read the documentation and troubleshoot guide

[dimpase]

@orlitzky @kiwifb

[saraedum]

@dimpase am I correct that this is only a blocker if #38163 gets merged?

[dimpase]

it is a blocker if any use is made of sagemath_environment in its current state.

[dimpase]

more precisely it could be used with another experimental package, but not with optional or standard.

[dimpase]

As you see, this bug (two bugs in one - dependency on an exterimental package + breaking --enable-system-side-packages)
has proliferated itself quite a bit

$ git grep sagemath_environment build/pkgs/
build/pkgs/sagemath_bliss/dependencies: bliss cysignals | $(PYTHON_TOOLCHAIN) sage_setup sage_conf sagemath_environment cython pkgconfig $(PYTHON)
build/pkgs/sagemath_categories/dependencies: sagemath_objects | $(PYTHON_TOOLCHAIN) sagemath_environment sage_setup cython pkgconfig python_build $(PYTHON)
build/pkgs/sagemath_coxeter3/dependencies: coxeter3 | $(PYTHON_TOOLCHAIN) sage_setup sagemath_environment cython pkgconfig $(PYTHON)
build/pkgs/sagemath_environment/spkg-src:cd build/pkgs/sagemath_environment
build/pkgs/sagemath_meataxe/dependencies: meataxe cysignals | $(PYTHON_TOOLCHAIN) sage_setup sagemath_environment cython pkgconfig $(PYTHON)
build/pkgs/sagemath_objects/dependencies:FORCE  cysignals gmpy2 | $(PYTHON_TOOLCHAIN) sagemath_environment sage_setup cython pkgconfig python_build $(PYTHON)
build/pkgs/sagemath_repl/dependencies: sagemath_objects sagemath_environment ipython ipywidgets | $(PYTHON_TOOLCHAIN) python_build $(PYTHON)
build/pkgs/sagemath_sirocco/dependencies: sirocco cypari cysignals mpfr | $(PYTHON_TOOLCHAIN) sage_setup sagemath_environment cython pkgconfig $(PYTHON)
build/pkgs/sagemath_tdlib/dependencies: tdlib cysignals boost_cropped | $(PYTHON_TOOLCHAIN) sage_setup sagemath_environment cython pkgconfig $(PYTHON)

of these, sagemath_objects, sagemath_repl, sagemath_categories are experimental. So these don't violate anything,
they are just broken, wrt to the 2nd bug.

For the other ones, something needs to be done. The most natural is to fix sagemath_environment so that
the 2nd bug is fixed, and promote it to optional, removing the 1st bug.

[dimpase]

I misread the diff for #38163 - the breaking change is much less recent.
This certainly does not make the situation any better, just shifts the blame from one PR to another PR, #35661

[dimpase]

@dimpase am I correct that this is only a blocker if #38163 gets merged?

no, this is not correct, apologies. The breaking changes came from #35661 - @kiwifb : you didn't notice there that it made a number of optional packages dependent upon an experimental package, sagemath_environment.

And it has broken --enable-system-side-packages