Conversation
Simplify logic of tests in NextDecimal_Tests to only test what is supposed to be tested. Should fix #2621
|
Well at least now it's consistent .... |
|
30 000 tests on my dev box, cannot reproduce. FML. |
For instance the `plugins` group was failing if tested on its own
|
So, still haven't found the specific issue with these two tests. I've improved things here and there in the process of searching I think, so in the end if we drop this PR make sure we still cherry pick useful commits. |
Still not fixing the problem but well.
|
So. I'm out of ideas. Still no clue why this happens sometimes, and not being able to reproduce on my machine despite the thousands of tests doesn't help. I think I'm just going to remove these tests and live with it :-) |
Don't fail if `phpunit --repeat 2`
…o issue2621again
FFS.
|
More random inconsistency with Yet this test is absolutely DB or option or whatever independent. Djeezus. |
This reverts commit b2f2be2.
|
Meeeeh. I think I've found. |
|
What a suspens! What was the cause? |
It would suck to have one day a stupid conflict, right? https://github.com/search?l=PHP&q=ut_print&type=Code
Duuuuuuuh. Of feckin' course using real plugins will modify YOURLS behavior, hence will affect unit tests... dumb.jpg
|
@LeoColomb testing the next shorturl id was failing when the previously loaded plugin was "random-shorturl". Captain Obvious to the rescue. The whole "let's load a real plugin to test the plugin API" was so dumb it's appalling in retrospective :) |
|
Nice one! 🎉 👍 |
Simplify logic of tests in NextDecimal_Tests to only test what is supposed to be tested.
(Replaces #2626 because I'm a git noob)
Should fix #2621