<@U3HLSKE8L> Now think about how Scala users feel.
# randomc
cedric
01/11/2018, 4:55 AM@adam-mcneilly Now think about how Scala users feel.
😂 5
o
orangy
01/11/2018, 10:43 AM
I’ve heard many times that last years this has been dramatically improved. Anybody checked Scala compilation speed recently?
d
dh44t
01/11/2018, 1:35 PMdaily
dh44t
01/11/2018, 1:35 PMit improves but not dramatically
dh44t
01/11/2018, 1:36 PMit also happens that sbt caches more things and reuse a compilation server
dh44t
01/11/2018, 1:37 PMthe overall process is faster if you take sbt on account
dh44t
01/11/2018, 1:37 PMbut isn't that fast
c
codeslubber
01/14/2018, 6:15 AMit's not near as bad as it used to be, tests run very fast inside the ide.. sbt still takes a long time to startup but apparently they are finally going to deal with that..
c
cedric
01/14/2018, 6:26 AMThey've been saying that for years. Both about sbt and Scala 🙂
d
dh44t
01/14/2018, 1:21 PMSolution for SBT is to put it to sleep and work on something else, Li Haoyi is working on a new tool https://github.com/lihaoyi/mill
👍 1
c
codeslubber
01/14/2018, 3:23 PMYeah I agree on that. Li's article about SBT was one of the best fishings of any ingrained tech tool I've ever read. I particularly loved the combination of there being an anemic tabularized model (rather than a tree/graph) and the fact that this allowed permutations that were senseless. Will be interesting to see how far he gets.
codeslubber
01/14/2018, 3:29 PMAs to what they have been saying for years, tests running in a few seconds is far and away the most important speed metric. Eclipse was a decade in and the only way they managed it was to have the maven plugin be braindead, and require frequent reboots (as it were). When that guy Fred came in and fixed the maven plugin, test times exploded (that was the end for me). If you drop to the sbt terminal in IJ and type test, it does nothing for the first 15-30s. But the tests run in the IDE very quickly, and never need to restart them. The sbt team might not end up fixing the startup issue, but then again, the JUnit guys all the sudden started making changes they refused to make as soon as TestNG showed up. But I agree with Mario because I think Li's article basically shows that sbt is totalled. So better to let a fresh approach rebuild it than the guys who made peace with the mess attempt to repour the foundation while they jack up what's there.