-
Notifications
You must be signed in to change notification settings - Fork 552
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
Cucumber tests failing in main branch #925
Comments
GRML... I have no idea right now and not too much OSS time on my hand. I'lll take a look hopefully at the latest on Saturday. It's just fun because this is a classic case of "works on my machine" which isn't too surprising (-CI) as that's probably how it got there in the first place. Just some... extra fun for me. Yay.
|
Also works on my other machine...
SO MUCH FUN |
Looks like the visibility checking isn't working on the CI and on your machine... summary_text = find("#{extra_class}.t-#{coverage_type}-summary", visible: true).text I checked that in the HTML for this one:
It's visible exactly once but somehow on CI and your machines it doesn't care about that/do the visibility checking correctly. So... I mean it could be race condition or maybe the css doesn't work as it should or what not... Also tried a fresh clone, all works for me. |
deactivating the
|
I'm starting to think that the JS is probably not working and I'm not sure why....
At this point maybe the assets are messed up or the the driver doesn't execute JS... I'm not sure. I confirmed google chrome is installed and apparition is configured... let's see. I think I'm taking a break for now |
Just a little note: I see that CI uses ubuntu-latest and I guess it got updated on October, 22 (see https://wiki.ubuntu.com/Releases). |
Good thinking but October 22 hasn't passed yet and ubuntu-latest refers to 18.04 apparently even 🤔 tried explicit 20.04 and 18.04, nothing works. At this point rewriting the class selectors to be unique among visible/not visible might be the safest bet to long term resolve this but that's quite some work I'm not quite sure I got the head space for right now :| no idea why when or how it happened... there were a couple of failures on |
Oops, I somehow messed October and August :( Sorry about that. |
you're on the right track though, theoretically it could have been only some package update but which one, why or how is... fun to debug with me being unable to reproduce. I could try the work mac for another data point... |
Yeah, must be some update. Maybe Chrome? |
Yeah... I can't quite figure it out. It seems somewhat most likely as that's one of the places where we differ as I'm not using chrome but chromium, difference shouldn't be big but still... :| |
well at least i can reproduce now on a machine... no time to fix right now the puma error is local, but the local duplications also happen... yayy macs
|
grml I thought I checked issues but now upon re checking I find twalpole/apparition#64 which is the likely cause |
Ha! I realize I might've run into the same thing in other repository, but I reverted the base docker image to the last working version and didn't dig further. I guess an easy test we can do is revert b0d8826 and upgrade to the latest cuprite to see if the issues I had with it when migrating to Github Actions are now fixed. That would be the third switch we make because of issues with the alternative 😆. |
😁 |
Hi, a little update on the issue. Currently things got better, but I still get 1 test randomly failing in main branch: https://github.com/umbrellio/simplecov/runs/1148966104?check_suite_focus=true Note that this is not specific to jruby since it also sometimes fail in MRI as well :( |
This is... very weird. Unless the error message is wrong:
If you look at the code: simplecov/features/step_definitions/html_steps.rb Lines 24 to 32 in c1a7e7d
this means that |
Actually expectation is on the wrong side there, it should look like this: expect(available_source_files.count).to eq(expected_files.length) I already fixed that in my branch since it's really confusing :) |
ah right the thing in expect is the actual not the expected... yeah that makes more sense. I hope I can give it another look this evening. It looks a bit scary as that might be further visibility shenanigans like I was facing with other browsers. Hopefully it's just a timing issue which I should be able to fix |
It's interesting... because we see the groups for one of the 2 examples but then don't see the source files... which is very strange.
|
In the mean time this seems to have been resolved.. mostly |
Some of the tests are still semi randomly failing like this one:
|
#939 begrudgingly itnroduces a retry to ward this off. |
At least this makes it possible to submit a PR, thanks! |
For some reason cucumber tests are failing for me when I run them locally in main branch. For example, I currently have a clean local copy on commit 9cd3a46 and here is what I get:
It looks like I have wrong version of simplecov-html, but it's locked to 0.12.2 in the Gemfile.
Build fails on Github as well: https://github.com/umbrellio/simplecov/runs/1063083286 (I just pushed the main branch to the forked repo).
Any ideas how to fix that?
The text was updated successfully, but these errors were encountered: