Clojurians Log v2

@mkainula might become able to test Spark 2 when we get 0.5.0 to clojars :)

Morning! I was just reading your weekend commits

What about

(defmacro with-resources
  "Setup resources and tear them down after running body.
  Takes a function, which when called, will setup necessary resources,
  and returns a function, which when called, will tear the resources down"
  [setups & body]
  (if-some [[setup & setups] (seq setups)]
    `(let [teardown# (~setup)]
       (try
         (with-resources [~@setups] ~@body)
         (finally (teardown#))))
    `(do ~@body)))

?

Otherwise, I’m ok to merge your changes. What should I do on the Github/Circle CI integration front?

Ok I believe I’ve done the integration part (basically nothing to do apart from authorizing Circle CI and selecting the right project).

thinking that powderkeg.repl might be simplifield if one important thing ia to have DynamicClassloader in place in the thread that runs in the driver

circleci branch does not pass currently, need to figure out more docker

Ok so not blocking on you for 0.5

with-resource was for having separate fixture per test ( spark 1.5 and 2)

yeah I understood

yep, 0.5.0 can go without circle

on simplifyng repl.clj: what do you want to simplify away?

may not need clojure.main

but not necessary to change right now

now refactoring kitchen since eating with my youngest one :)

Forgot to mention it but 0.5.0 is on Clojars, @mkainula enjoy 😉

@cgrand so now had time to read with-resources with a thought, your version looks neat 🙂

so this would allow to do (with-resource [setup-1 setup-2 setup-3] (stuff that matters for testing here))

got the idea for with-resource after reading this: https://stuartsierra.com/2016/05/19/fixtures-as-caches

hum, cannot get spark 1.5 test to work, worker and master don’t manage to talk to each other 😕

ok, seems that master and worker talk to each other but connection to master fails, need to figure out what happens in the containers

Do you know if it may be an interference with spark2 tests having been run right before?

that actually shows in circle as reusing name given to container, since it doesn’t remove old containers (or remember reading something like that)

but connection problem persists when running single test locally

well, have to go to sleep now, might look into this another day

anyways, I guess the core development would be better headed to DataFrames/DataSets, this testing thing is a neat sidetrack 🙂

@esma