From MozillaWiki
Jump to: navigation, search


digitarald, dreams, jgraham, maja_zf, mmucci



Dashboard (digitarald)

  • Revive old puppeteer-firefox dashboard? Use test metadata from James' patch above?
    • Grab the JSON data from central and break it down by test.
  • On TaskCluster we run (almost) all tests, failing and passing (in puppeteer ci tests failing tests disabled, no way to track test status metadata)
  • jgraham can help with infrastructure
  • digitarald: How do we communicate the MVP scope in terms that are useful to Puppeteer users?
    • jgraham: How to know which things don't work and which are expected to work?
    • maja_zf: Disable irrelevant tests as out of scope?
    • jgraham: But we don't want to imply that we absolutely won't implement something; lose out on early feedback.
  • dreams: Narrative of where did we go deep
    • Go deep on a few key areas that will be helpful to most users

digitarald: How can Puppeteer tests be improved to help Firefox's productivity

  • Test harness is not robust. Lacking errors for failures and timeouts.
  • Tests are not isolated well in terms of API usage.
  • Should we run CDP tests, and which?
    • Do we get more detailed feedback on what's wrong?
    • maja_zf: Running them seems impractical since they are spread all over Chromium code base, tightly coupled with Chromium components. Amounts to much more work than just importing a test directory into central.