Open cornelius opened 5 years ago
This would be a great improvement!
This setup would also run the linting in parallel with travis builds, and travis build's would not have to do linting. That way a linting check may still fail while also telling that the build itself is alright (with travis right now the build won't even be executed).
Having a bot would enable us to just enable it for new repositories instead of copying linters, etc.
A quick look at GitHub apps:
There is Hound (home page) which does much more than just code style checks. It does support Python, Javascript, Typescript and some other languages but not C++.
There is CodeFactor (home page). It also does much more than code review. It's not clear from the home pages how well it does linting. It does support C++, Python, Javascript and Typescript.
There is probot/linter which is a simple Python linter.
There also is lint-review, which is a bot but not available on the GitHub marketplace. It's open source so you can run it yourself. There is a hosted version as Stickler. It does Python, Javascript, TypeScript, but not C++.
Writing own bots for GitHub Apps is made relatively easy with Probot.
We currently run linting in Travis as part of the tests. It would be better to run linters separately as bots, e.g. as GitHub Apps. This way they could easily be shared across all repos, they would give early feedback on pull requests, and they would not interfere with other tests.