Now that I see this, I'm a bit concerned about https://github.com/easybuilders/easybuild-framework/blob/develop/easybuild/framework/easyblock.py#L1405
Are there cases (such as a bundle) where this will come back to bite us?

• Failed build of Scipy-bundle using current develop
  https://gist.github.com/lexming/5808a464c1a38771fd9d697df71f9d6c
• Successful build of Scipy-bundle using this PR
  https://gist.github.com/lexming/53fc1fb229630f6c1eda0ebeb996e099

• Successful build of matplotlib v3.2.1 using this PR
  https://gist.github.com/lexming/3dc52f3a0b51a91add1591b221fad29c
• Successful build of IPython v7.9.0 using this PR
  https://gist.github.com/lexming/b4dde86e2d6cbef9d94645d54037eb50

Now that I see this, I'm a bit concerned about https://github.com/easybuilders/easybuild-framework/blob/develop/easybuild/framework/easyblock.py#L1405
Are there cases (such as a bundle) where this will come back to bite us?

If that's a problem then it should make the sanity check fail due to missing stuff in the module file?

In this particular case, we're only making sure that the module file that is used to set up the environment before installing extensions is correct.

The module file used during the sanity check is a fresh one, which should have the same contents as the final module file (since nothing is changed anymore when the sanity check kicks off, that's after (post)install and extensions steps).

Fix confirmed with eb SciPy-bundle-2020.03-foss-2020a-Python-3.8.2.eb -f --include-easyblocks-from-pr 2081 after committing @lexming's suggestion, so this should be good to go...

Going in, thanks @boegel !