-
-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ReaderHighlight: unit test #11775
ReaderHighlight: unit test #11775
Conversation
Can I trigger the test without merging? |
If you can't run the tests locally you can toss in a make coverage or something. Line 12 in 8530282
This sounds extremely unlikely though. :-) There are some that say #notest. Those might fail in CI for various reasons, so they're only for local. Some of those should be reevaluated now that we're a few years on. Many more say #nocov for similar reasons. However, the tests are split up over 2 jobs to finish quicker while the coverage does all of them. |
Concretely: job 0:
job 1:
= 289 + 250 |
I must say, I don't think I ever managed to run the frontend testsuite successfully on master: I always get some failures in the readerhighlight part (and the occasional segfault). I have never reported the issue because it works fine on my meson branch, where I've made some changes to speedup the testsuite and I don't have to fiddle with the environment (luarocks, busted, etc…). ¯\(ツ)/¯ 2 commits touch |
Can't confirm, unless it somehow changed in the last few months. Normally I only run specific tests locally. (Unless you're referring to Turbo or coverage.) I don't know why that was waiting a whole second but based purely on that being there it would suggest |
Why, too slow? Anyway, it might fixed be another change to another part of the testsuite, as there's some sort of interaction between the tests. On master (8530282):
(disregard
|
Obviously, but unless you can make all tests run in under a second on a slow laptop I don't see any scenario where that would ever change. (To be clear and for context I made Travis/CircleCI more than 10 minutes faster than it once was.) If I'm working on ReaderHighlight there's no point in wasting time on anything else. The automated equivalent to that manual action, limiting tests based on changed files the way I did in crengine seems far too risky, but perhaps it could be useful as a development aid. |
That much? Impressive, that's better than I had anticipated. I didn't think there was much more than about 30% to be squeezed out of it. (On CI I mean, of course there are easier gains to be made with more cores.) Though like I implied, the difference between 10 seconds and 100 seconds is barely perceptible to me. They're both too long, let's go get some tea. ;-) |
It's mostly thanks to parallelization:
Less resources used on the CI actually, but better used: no parallel runs (example run): |
On master, I can't run the Fixed by c7b31e2. I can now run the full testsuite with #11784. But there are still some rotten things in the busted kingdom. For example, I still get 1 failure when running |
Yes, some tests may still have issues like that. |
Fixed by #11784 |
Thank you. |
No idea of the reasons of #11563 (comment) so far.
This change is