Remove RHEL-8 workarounds from old jsonschema

[mcasquer]

As tmt running in RHEL 8 is no longer supported, remove the schema workarounds referencing RHEL-8

Fixes #3705

• modify the json schema

[psss]

/packit build

[LecrisUT]

I had my suspicion about this comment, and I tried it with the following example:

/tests:
  /valid:
    test: something
    require:
      - url: https://github.com/beakerlib/beakerlib
        type: library
        name: /example
  /invalid:
    test: something
    require:
      - url: 12345
        type: library
        name: { }

With the changes here, the invalid one always passes the schema, while previously it failed correctly. Best option here is to just delete these comments and revert the rest.

[mcasquer]

So the options such as url: true are not valid, we still need to redefine them, right?

[LecrisUT]

Yes, the previous way of defining them was correct. Maybe just needing more $ref indirections.

[LecrisUT]

Make sure this comment is preserved, it is still very much relevant

[mcasquer]

Done :)

[mcasquer]

What about the tmp_path* fixtures that are not available at RHEL-8, I see we have a workaround at

tmt/tests/unit/conftest.py

Line 45 in 746105f

class TempPathFactory:

I'm not sure of the magnitude of what this implies, or if it's good as it is now :)
cc @LecrisUT @happz

[LecrisUT]

What about the tmp_path* fixtures that are not available at RHEL-8, I see we have a workaround at

tmt/tests/unit/conftest.py

Line 45 in 746105f

class TempPathFactory:

I'm not sure of the magnitude of what this implies, or if it's good as it is now :) cc @LecrisUT @happz

Based on

tmt/tests/unit/conftest.py

Lines 39 to 44 in 746105f

	# * a wrapper class, representing the "tmp path factory". It's initialized with an
	# actual factory, has the same public API, but returns our `tmt.utils.Path`
	# * two new fixtures, `tmppath` and `tmppath_factory` that consume available fixtures
	# and return corresponding `tmt.utils.Path`.
	# * tests using `tmppath*` instead of pytest's own `tmp_path*` and `tmpdir*` fixtures
	#

We still need to keep it, but the try ... except in there can be simplified. What I don't know if it was invesitgated, what happens if we also provide a pytest.fixture with the same name tmp_path or have any way to keep the tmp_path name but edit the output type.

But the minimum blast radius would be to just simplify the try ... except and that should not have any side-effects, so start from there.

[happz]

What about the tmp_path* fixtures that are not available at RHEL-8, I see we have a workaround at

tmt/tests/unit/conftest.py

Line 45 in 746105f

class TempPathFactory:

I'm not sure of the magnitude of what this implies, or if it's good as it is now :) cc @LecrisUT @happz

If you keep tmppath and tmppath_factory working as they are today, and drop the backward compatibility, the it should be harmful and remove one old piece of code.