It's a great step towards the right direction, certainly! I feel the deriv. times could be addressed in the compiler more systematically.
-
-
Replying to @jvican @ScalaIO_FR
There's not much more the compiler can do: My ASTs are semantically quite minimal yet most of the compile time is still typechecking them.
1 reply 0 retweets 0 likes -
The compiler may help with the implicit search and AST generation, but those ASTs are fundamentally large and therefore slow to typecheck.
1 reply 0 retweets 0 likes -
Replying to @propensive @ScalaIO_FR
I would need to have a closer look, but I can tell you I'm tired of benchmark the typer and 90% of time is in implicit search, no asts.
1 reply 0 retweets 0 likes -
Generally, simple AST typechecking is really cheap. The cost of implicit search dominates compile times in shapeless and Circe.
2 replies 0 retweets 0 likes -
Replying to @jvican @ScalaIO_FR
I also cache recursive implicit searches for the same type, though in my tests, that didn't buy me much.
1 reply 0 retweets 0 likes -
Replying to @propensive @ScalaIO_FR
I said that *simple* AST typechecking of big trees is not a problem in all the benchmarking I've done so far :) I can share data if you want
2 replies 0 retweets 1 like -
I would be interested in having a closer look to Magnolia compile times, I'll benchmark it today with new stats. Is it 2.12?
1 reply 0 retweets 0 likes -
Replying to @jvican @ScalaIO_FR
Yes, that would be interesting. You should be able to clone `github/propensive/magnolia` and run `sbt tests/compile`...
1 reply 0 retweets 0 likes -
But you'll need to also clone `http://gitub.com/propensive/estrapade …` and `sbt publishLocal` on that first, or just remove any references to Estrapade.
1 reply 0 retweets 0 likes
(Or alternatively, wait a couple of days until I've published both.)
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.