Skip to content
Tim Watson edited this page May 20, 2014 · 1 revision

Now we're not constrained to do anything in any particular way, let's have a rethink… Here are some initial steps that could take place:

  • disambiguate testing from monitoring and decide on a niche
  • use the "big refactor" thing to get out of managing only Erlang nodes
  • think about how we present the data that we're capturing

and also, at the moment (and with future plans in mind) what makes us different from

  • common_test/test_server for Erlang
  • HPQC / OATS / etc for anything else
  • RUEI and various existing things that produce monitoring info

The missing "bit of the puzzle" isn't really about testing or monitoring IMO, but goes back to https://github.com/hyperthunk/axiom - it's about testing integrations, where those happen to form a distributed system.

On that note…

  • we need support for heterogeneous interfaces
  • it would be nice to think about multiple ways to feed test data on ingress
  • and likewise we should be flexible about egress
  • making assertions ought to be a declarative thing
  • we shouldn't re-invent the wheel (i.e., attempt to be a test framework)
  • we should find ways to work with existing wheels though

So, on the various interfaces front, Erlang's ecosystem isn't actually as feature rich as Java's right now, though it's not far off.