Replies: 2 comments 2 replies
-
Sounds good.
Most of the test failures are related to the infra, not to the tests themself. In general, the test that checks if we bytecompile is the most useful test in this repo.
I don't think having policies like that will be helpful - a lot of the contributors are zero/no elisp knowledge but still the PRs they are adding are useful for the community. Also, a lot of the features are not easy to test or at least they will require a lot of work. E. g. consider adding a server setting. If we go this way we will end up with a test suite that is slow and red all the time. If we have 2 active contributors per server - then fine, we can split the project in multiple repos and then let each team build their suite. But in reality, we are even struggling to cover the spec, sync up the server settings, and so on. On the positive side, I don't think that we have a lot of regressions - we used to have a lot more. |
Beta Was this translation helpful? Give feedback.
-
I would recommend Eask, it's very similar to Cask. The previous PR is in #3465. I closed the PR due to some ert tests I am not able to resolve myself.
Does that mean we can remove some of these tests, and re-implement the new one? 😕 |
Beta Was this translation helpful? Give feedback.
-
As more and more clients are added and the project grows, bugs are bound to appear. So it only makes sense to try and mitigate these by adding more tests.
The current suite is noticably brittle and difficult to manage, and while I would undertake it myself; it makes more sense to first discuss some details and come up with some standards to make future usage much smoother.
And anything else I might have missed!
Beta Was this translation helpful? Give feedback.
All reactions