On Mon, Jan 8, 2018 at 3:07 PM, Åke Sandgren <ake.sandg...@hpc2n.umu.se> wrote:
> > > On 01/08/2018 01:58 PM, Caspar van Leeuwen wrote: > > I disagree, because the requirement that the variable 'BAR_BIN' that is > set is a requirement from Foo, not a 'feature' from Bar. Bar knows > absolutely nothing about this variable, and is very happy to run without it > - thus it has no place in a Bar module. To illustrate that more: what if > someone tomorrow develops a Foo2, which requires another environment > variable 'WHERE_CAN_FOO2_FIND_BAR' to point to the Bar executable, would > you reinstall Bar and put this additional variable also in the module file > of Bar, just because Foo2 needs it? > > Yes, I would re-make the module file for Bar with that env added. > > If FOO needed "PYTHON_BINARY", I don't think FOO's poor decisions should be allowed to pollute the Python module. So then FOOs easyconfig couldn't make it upstream (unless it relied on everyone doing manual local modifications). If BAR is almost always used in conjunction with FOO, and there might be user compiled versions of FOO that depend on BAR and need the same environment variable etc, then the situation would be different.
