[Bro-Dev] [Bro-Commits] [git/bro] master: Updating some test baselines. (827dcea)

[Jonathan Siwek]

>> Changing things in advance without actually running the tests to make
>> sure it works as expected seems dangerous.
> 
> That's not what I wanted to say. :) Of course one still needs to run
> the tests before pushing things further upstream.

Ah, ok.

> But it's easier to first adapt the tests and then run them

Maybe it's just more my style, but I'd rather see what the failure looks like first before adapting anything.  Otherwise I might be adapting something that didn't need adapting or adapting it in the wrong way such that I'll still have to fix it later.

I'm still thinking that a NOTES file that maintains a list of tests that are expected to fail more often because they're coverage-type tests is going become out-of-date easily.  Supporting evidence: I just noticed that the testing/btest/README is in need of significant updating because it tried to maintain a list of major test directories.

Right now all the tests that are expected or more likely to break when new code is added have "coverage" somewhere in the name.  What do you think about grouping them all in a single "coverage" directory to keep that distinction, but then also provide a canonical place for preemptive test-fixers to look?

- Jon