re-learning transducers (maybe I didn’t really learn them when they appeared :)), watching Rich’s strangeloop talk https://www.youtube.com/watch?v=6mTbuzafcII and discovered slides of Christophe’s ClojureD talk https://cdn.rawgit.com/cgrand/xforms/resources/Lost%20in%20Transduction.pdf

@viesti re:cider it takes a loooong time or never finishes or fails?

didn’t have the patience to wait 🙂

should look at the dependencies that get pulled in, maybe some of them could be filtered out

I believe counterclockwise embeds cider and while I don't see 10k classes it's way slower than plain lein repl. For about the same number of classes.

do all the classes on the classpath get instrumented or just the ones that belong to the project using them?

thinking if scope of work could be narrowed

hum, had [refactor-nrepl "2.2.0"] in addition to [cider/cider-nrepl "0.14.0"] in my leiningen profile

I think it's a different issue as with recent counterclockwise I experienced nearly a 10x slowdown while the classes count stay almost identical.

However looking at https://docs.oracle.com/javase/7/docs/api/java/lang/instrument/Instrumentation.html again I see a couple of ways of making ouroboros startup less brutal.

that would probably be neat, although nowadays I have cider in a separate profile that I enable only when needed to avoid startup time cost which it brings

hum, was thinking if SparkSession (of Spark SQL http://spark.apache.org/docs/latest/sql-programming-guide.html#starting-point-sparksession) should be wrapped somehow, at least for making Spark's own examples easier to translate to powderkeg

but browsing through the code of SparkSession, it looks more like a wrapper class for Spark’s own API evolution, which might not be our target

another thing that I’m thinking is should we be worried about Encoders or not 🙂 https://spark.apache.org/docs/latest/api/scala/index.html#org.apache.spark.sql.Dataset

Yeah not very open.