Open
Description
Is there an existing issue for this?
- I have searched the existing issues
Code improvement description
❓ We get regular bumps of the linters/static code analyzers used in native helpers like
eslint
, phpstan
, etc, and even though it's relatively trivial to bump them, it doesn't add much value to do that week after week. A few options I see:
- keep doing it and eat the cost.... it's not much focused time, the more annoying part is the distraction of having to remember to
@dependabot rebase
the next one since we aren't as aggressive about rebasing anymore - add merge-queue on
dependabot-
core... it's been helpful for this use case in an internal repo, for safety I only queue up one per ecosystem to avoid stepping on each other, but this would let us merge one-linter-per-ecosystem concurrently across multiple ecosystems - wait for grouped-updates to land... this only solves a small part of this particular problem though
- migrate them one-by-one to using a GitHub action to handle linting... we actually used to do this (GitHub Action: Add Python flake8 linting #2892), and then switched to embedding (breadcrumbs in CI: Simplify workflow by moving suite specific tests into test script #3430)... but embedding isn't necessarily needed now that many editors such as VS Code and others allow quickly installing language-specific plugins like linters...
- migrate them en-masse to using https://github.com/oxsecurity/megalinter... again a CI thing, but keeps some complexity down because it "Just works" on all the native helpers, w/o us having to configure them... eg, we need to add
yamllint
(Runyamllint
on PR's #5572) and haven't gotten to it, this would let us pick that up for free
One tricky bit is illustrated by #6830. Some linters will need to be run against a matrix of runtime versions.
Thoughts?