How / whether to handle tests that come with some software
Opened this issue · 2 comments
Often the source for some piece of software comes with a set of tests of some sort. It can be meaningful to execute these tests to get some evidence that the build results function appropriately.
Running tests takes time + resources and sometimes the tests depend on things that are not necessary for building or executing the software.
It is unclear what to do about these types of tests.
I would love to be able to choose per package whether to build & run tests (incl. all the extra deps) or not. This is kinda how it works in Nix now.
I encountered this text about ncurses at LFS:
This package has a test suite, but it can only be run after the package has been installed. The tests reside in the test/ directory. See the README file in that directory for further details.
I found that surprising -- my memories of testing things I've just compiled are before running something like make install. May be this doesn't happen too often.