Open gareth opened 6 years ago
Hey @gareth, definitely! We are working on it and should be available in the Marketplace very soon.
What are you most excited about with the new Checks feature? Are you using any other services that already use Checks, like Travis? We'd love to take a look at how it's being used in the wild.
If I'm honest I don't know the ins and outs of the new API, I'm waiting to be shown what great things can be done with it. Given the impact of the change I'm assuming there are some significant benefits I haven't realised yet.
For Hound I guess not getting duplicate comments for PRs with multiple commits would be a big advantage. Obviously being able to see the linting results as their own separate result would be great. Plus any change that means I can limit the permissions I have to give integrations is a bonus.
I would also like to see a feature, where all code style violations are available through houndci.com, not via comments. With big PRs comments become very messy very quickly. Another thing is, that in current implementation (at least with eslint) Hound always passes, even though there are code violations. In my opinion it should fail, giving indication, that not all checks have succeeded.
Another thing is, that in current implementation (at least with eslint) Hound always passes, even though there are code violations. In my opinion it should fail, giving indication, that not all checks have succeeded.
This is already possible by setting fail_on_violations: true
in your .hound.yml
file, as documented in our configuration
.
@gylaz Awesome! Didn't know that, thanks :)
Big 👍 to being able to view all violations on houndci.com rather than PR comments. Ideally the "N violations found" check on GitHub would link to the appropriate page.
@salbertson Our use-case is simply to keep the feedback by humans separate from the feedback by the dog. If Hound inserts 10 comments in a file it's hard to see the actual code. But temporarily turning off comments, would also hide the human feedback entirely.
So this has to do with cleanness and thus work efficiency for us :) Thank you for implementing this (I hope though that this issue will not be open until 2020 or so ;)
@mmzoo love it! Thanks for the feedback, we appreciate it.
Is anyone interested in helping us out with this?
@salbertson I have Friday time at work that I could contribute. I would be interested in contributing to an option that keeps Hound feedback separate from human feedback.
@salbertson Back in May you mentioned that this was in progress. Anything I can work off of or would it be better to start from scratch?
@DylanReile that work was mostly adding support for our GitHub App so we could start using Checks. Starting fresh would probably be best.
I just noted another disadvantage with comments: they don't disappear when they are outdated. They just hang around in the main conversation even though they have been fixed:
Unfortunately I bit off more than I had time to chew by volunteering for this. At the time I assumed the architecture was something like this: PR -> Hound webhook -> Run checks, assemble a big array of review comments -> batch submit comments on the PR
I further assumed I could just swap out the last batch commenting step with a batch checks uploading step. This was naive. The existing architecture actually streams comments as they become available; not all at once.
Moreover, the checks API is more sophisticated than PR -> trigger checks. There are events like pausing checks and re-running checks.
Fitting these complications into the existing control flow is more than I had time for. Sorry about that.
This is now needed for the built in GitHub Reminders system.
Without it, we can't get notifications of failures. ie, the entries in the bottom field do nothing.
Is it on the roadmap at all?
Hi!
We love Hound, and were wondering if there were plans to integrate it with Github's new "Checks" API (and I guess, make Hound accessible via Github Marketplace)?