On Nov 22, 2020, at 07:55, Jonas Hahnfeld <[email protected]> wrote:
>
> . . . had some concerns about Spanners that were dismissed on
> the basis that there was no test case that showed wrong results (is
> that really our standard for claiming that a change works correctly?!).
To be fair, there are things in this program that are complicated enough that
it's unclear how to reach a specific state. On the other hand, there are areas
where existing tests cover the basics but give less attention to boundary cases
and interactions between features.
Demanding that a reviewer produce a test case is a faux pas where I come from
(midwest U.S. and Canada). A productive response would be something like this:
I spent half an hour trying to create a test case to address
your concern. The closest I got was . . . . That doesn't
quite cover it. Do you have any suggestions? If not, I'd
like to move forward with this patch.
Regards,
—
Dan
