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
Feedback about the new pull request comment #8
Comments
I think it's a great feature, but it can sometimes clog up a PR review screen, so it would be great if they were collapsible so that we can read the code around them more easily when desired. |
@TechMaz thank you for your feedback. Would love to hear your feedback 🙏 |
@eliatcodecov httpie/cli#1434 |
I love it! It's so much cleaner and less obtuse <3 |
I'm not sure how that's possible but pull request form shows me wrong coverage result - it says my pull request will decrease project coverage, although I'm confident my changes won't not affect coverage at all. I don't see that problem in the codecov web application. I also compared coverage.xml files uploaded to codecov for corresponding commits and they're identical. The pull request with wrong form: Full report at codecov: And here's the coverage files for corresponding commits:
|
Hi, thanks for providing codecov tests to the Open Source community for free. I do not trust the codecov report, because why it's going down when I literally create the first test file in this package? PS: I just measured the test coverage of the new file created, this is 87% (according to IntelliJ). |
This may sound like a nitpick but the coverage diff has a sign error in the GitHub comment that the bot is posting. It says "coverage decreases by |
On this pull request I simply added some lines to the documentation (README.md and README.rst), but CodeCov warned that code coverage had been reduced by my change. It would be nice if obvious documentation files didn't count. Thanks. |
Hi, On this PR, I just copy an old case and then mod one function call, no tests were removed, and the test coverage decreased. |
I'm unable to view the coverage report like before on the PR's. Codecov is such a big mess rn. |
Thank you, @KnorpelSenf that makes sense - will update text clarification in next iteration. |
|
Is it possible to disable the automatic comments? We'd like to keep using the Codecov GitHub app (so that our Update: it looks like comments can be disabled by adding |
Is it possible to manually request that codecov re-run the report? I've found the most spotty aspect of codecov in our development workflow to be when we force push to a branch with new test coverage but codecov doesn't update the statistics. It would be great to be able to basically say "@codecov rerun" or something along those lines in the PR. |
Hey! Over at https://github.com/Significant-Gravitas/Auto-GPT/, we are running into some severe problems with CodeCov. We tried reaching out to a rep and getting some of it worked out and a demo, but they were unavailable at our scheduled time. We liked CodeCov, but it currently needs fixing and denying all PRs. I don't particularly want to swap providers, but hundreds of people are asking us why CodeCov is broken, and we need help finding answers. I would love a reply as soon as you can |
What about the StateHasChanged implementation for the other components, like MudTabs and etc. ? |
The pull request comment seems good to me (I'm just starting with Codecov), but there seems to be a bootstrap problem which I describe here: when there is nothing to compare to (because the base commit doesn't have any code in it yet), it's treated as a error situation instead of as though all the changes are new code (which...they are 😁). Once my PR is merged, we'll presumably never encounter the situation again, so perhaps it's not real important, but fixing it would improve the user's on-boarding experience, and it doesn't seem like it would be that hard to address. |
@webbnh thank you for your feedback!
💯 that's right, the comment has a missing base report and looks like an error 🙉 . We are actively looking to have the 1st comment more welcoming and clear as seen below. If you have any thoughts or revisions you'd make let me know! |
Those are rounded numbers, so it could have gone down by 0.008 from 99.614 to 99.606 which would be rounded to 0.01 from 99.61 to 99.61. Perhaps we would want the diff to operate based on the rounded absolute values, but that could effectively hide a decrease in test coverage. |
Explicit rounding aside, floating point math is much harder than it seems. (Any time you see a decimal point in the middle of the number, assume that anything to the right of it is "an estimate"! 😁) |
@EverWinter23 thanks for the feedback, was hoping you could expand on it:
Do you have a screenshot of what your referring to? Any additional details can really help the team improve the format. Thank you 🙏 |
@donhcd can you provide a link to your PR or a commit SHA? |
Meta request: The way this solicitation of feedback is linked in the comment means that anything that syncs comments to places with previews (like Slack) results in a giant blob about this feedback request in those channels. If you made this go through a link shortener or such to break the preview, it would be really nice. |
The codecov report is bit inconsistent with on different runs for same test set. |
@naik-aakash can you be a little more specific about that? Which commits for example? Do you mean that the coverage results are different or that the PR comment format is different? Screenshots would be helpful as well. |
Yes, sure, sorry. Coverage results are different with multiple runs. I suspect it had to do with layout_dicts.py file formatting now. Not entirely sure that was root cause for this or not. Here is the PR where you can see the inconsistency. |
I'm not sure if this is only me reading this incorrectly, but I'm having a hard time truly understanding the difference between the PR comment and the report itself, they don't seem to be aligned: The number of missed lines is not the same. |
@HarelM thank you for the detailed report 🙏 |
Is there a way to make codecov not send the report when all is well? I really only want to hear about failures. Thanks! |
It's also worth noting the following experience: I know our use case might be complicated, but coverage is very important to our project, I've done a lot of tweaking to be able to measure coverage correctly. |
@dabrahams there is this configuration that will only show if ANY changes occur. That doesn't quite acheive your ask though. cc'ing @thomasrockhu-codecov @drazisil-codecov are you familiar with another configuration to resolve this? Created an issue to look at this more / adding ability to show only with negative reporting: codecov/engineering-team#1350; please add any thoughts/comments you have. |
@HarelM it may help to apply after_n_buid here, do you have this set? It will wait until that many reports are complete to update the comment. EDIT: adding issue to look at this further: codecov/engineering-team#1352 |
Ah, didn't know about this option, I'll try it out, thanks! |
@codecovdesign I don't see the specified flag in the github action parameters here: |
@HarelM if updating flags you'll want to do this in the codecov.yaml https://docs.codecov.com/docs/flags#step-2-flag-management-in-yaml |
@codecovdesign is there a way to do it without introducing a whole new file to my repo in order to send a single flag to codecov? Is it possible to support the most common flag (or better yet all of them) in the github action configuration? I'm managing most of my CI configuration in this file anyway, it would be great to be able to simply add |
There's a bug that's preventing from getting accurate report due to the fact that it takes too many files (not the files I specifically requested): |
@HarelM sorry missed this one; @drazisil-codecov and I were just looking at it and were unsure exactly. would you be up for a call? if so, shoot me an email at kyle.mann@sentry.io and will find a time. curious to learn more about your use case too. |
I've sent a mail to see when we can meet. |
In the go-github repo, we consistently see messages like:
without any explanation as to why this is (or how to fix it). As a result, the codecov report is not useful. |
Some time in the last few weeks, the reporting of coverage by lines with diffs stopped working, as in this PR: xdslproject/xdsl#2540 |
Thanks for dropping by! 👋
We've been iterating and updating the layout, summary, and copy of the pull request comment.
We greatly appreciate your time and thoughts - looking forward to hearing from you ❤
Codecov team
This issue is intended to share and collect feedback about the tool. If you have support needs or questions, please see our support page.
The text was updated successfully, but these errors were encountered: