On 17:59:46 2019-02-02, Jani Nikula wrote:
On Fri, 01 Feb 2019, Bruno Santos <brunomanuelsantos@xxxxxxxxxxxxxxxxxx>
wrote:
Test sources are not the ideal way to automatically discover tests
because every test requires a source file. This would require extra
metadata to discriminate which tests should be performed on each file,
which would be a needless maintenance burden.
Discovering tests based on the files used to check their success means
that sources can be easily shared across different test types and the
tests that are actually executed are in fact encoded in the file system
itself.
This scheme has the advantage of allowing test cases to be introduced in
a phased manner or opted out entirely in case they're somehow redundant.
All previous tests are still executed: so far only two types of tests
are supported and they're both related to the same .rst files.
Maybe I'm dense, but I don't quite grasp what's the advantage
here. Perhaps it would be useful to get a concrete example what the next
steps would be.
BR,
Jani.