by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 17:25
    adpi2 synchronize #790
  • 15:10
    dwijnand commented #744
  • 15:09
    dwijnand converted_to_draft #744
  • 15:04

    dwijnand on wip

    (compare)

  • 14:54
    dwijnand ready_for_review #744
  • 14:48

    dwijnand on wip

    Add pipelining to IncOptions Implement early output and earl… Expand CompileProgress and Setu… and 5 more (compare)

  • 05:28
    eed3si9n commented #790
  • 05:23
    typesafe-tools commented #791
  • 05:23

    eed3si9n on develop

    zinc-scripted: Remove indirecti… Merge pull request #791 from dw… (compare)

  • 05:23
    eed3si9n closed #791
  • Jun 03 10:17
    typesafe-tools commented #791
  • Jun 03 10:17
    dwijnand opened #791
  • Jun 02 18:20
    dwijnand review_requested #790
  • Jun 02 18:11
    dwijnand commented #790
  • Jun 02 17:00
    eatkins commented #790
  • Jun 02 15:17
    typesafe-tools commented #790
  • Jun 02 15:17
    adpi2 opened #790
  • Jun 02 09:31
    bishabosha commented #779
  • Jun 01 13:44
    dwijnand closed #306
  • Jun 01 13:44
    dwijnand commented #306
but–this util is an interface with a bunch of static methods. IMHO it makes more sense for it to just be a class...
eugene yokota
@eed3si9n
Zinc 1.3.0-M2 is on its way to Maven Central
Guillaume Martres
@smarter
sbt/zinc#637 :scream:
Guillaume Martres
@smarter
Also I don't know why you're checking scala versions here but this doesn't look very future-proof...: https://github.com/sbt/zinc/blob/694d60f9cc7b2b8b5d0c4cadde313ca727454f9e/zinc/src/main/scala/sbt/internal/inc/IncrementalCompilerImpl.scala#L257
is this supposed to be >= 2.12 ? From the flag that is passed I would guess yes. I guess the good news is it'll be fixed for a while when we get to scala 2.120
eugene yokota
@eed3si9n
is that passing -YdisableFlatCpCaching?
Guillaume Martres
@smarter
looks like it
and I'm not even sure why because the classpath caching is supposed to be aware of timestamps now so should be fine
eugene yokota
@eed3si9n
eugene yokota
@eed3si9n
JDK 11 testing is failing with seg fault
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007fdc0d6e5351, pid=109, tid=984
I wonder if the alpine image is missing some library that openjdk11 is using
eugene yokota
@eed3si9n
I got the tests to pass on JDK 11 - sbt/zinc#640
Jorge
@jvican
Great, I’ll have a look today then
eugene yokota
@eed3si9n
sbt/zinc#601 has been open for a while
it's a cleanup PR that removes most of warnings, but since it touches a bunch of files it conflicts easily
eugene yokota
@eed3si9n
just published zinc 1.3.0-M3
eugene yokota
@eed3si9n
@jvican do you know what's going on with https://ci.scala-lang.org/sbt/zinc? it can't seem to download artifacts from Maven Central
eugene yokota
@eed3si9n
ftr, CI is now back in operation
Piotr Kołaczkowski
@pkolaczk_twitter
How to publish zinc locally so I can test it with another project?
eugene yokota
@eed3si9n
open sbt, +publishLocal
eugene yokota
@eed3si9n
@/all please check this out. Lightbend Community Code of Conduct - https://discuss.lightbend.com/t/lightbend-community-code-of-conduct-for-all-repos-under-sbt/5099
eugene yokota
@eed3si9n
gave a talk on Zinc at ScalaSphere - https://twitter.com/eed3si9n/status/1181878462165983232
Dave Nicponski
@virusdave
heyas, qq: Is there, or has there ever been a zinc issue using classpath manifest jars?
eugene yokota
@eed3si9n
I'm not aware of it but I won't be surprised if the incremental compilation doesn't account for Class-Path entry in MANIFEST.MF
Dave Nicponski
@virusdave
hmm :/
Initial attempt makes me think it's not even working for clean build. Will investigate more today
Donald Andrew Bronson
@Recon419A
Hey, quick question. I pulled sbt/zinc and the version of zinc in the build.sbt is 1.3.0-SNAPSHOT while the release version is 1.3.3. Is there a reason the build anteversions the release?
eugene yokota
@eed3si9n
no that should be 1.4.0-SNAPSHOT instead
Donald Andrew Bronson
@Recon419A
Okay, just wanted to be sure I wasn't smoking something.
eugene yokota
@eed3si9n
Alex Shubert
@lure
Hi everyone. I tried to upgrade to Gradle 6 from Gradle 5 and build time increased two times.
The apache/kafka#7677 states, this is a tradeof in new zinc
The new version of Gradle uses a new version of Zinc which trades full compilation speed for incremental compilation speed.
However can't find the explanation on that in zinc docs. Is that really the case? Are there any worksrounds. Thank you.
Guillaume Martres
@smarter
no, that's really weird/unexpected
seems like a gradle bug
Guillaume Martres
@smarter
zinc shouldn't be adding more than a few percent to compilation time
a sudden 2x slowdown means something has gone terribly wrong somewhere
Alex Shubert
@lure
It is consistently two times slower. Just replacing 5.6.2 to 6.3/6.2.1 and back provides stable 11m vs 6m. Nothing else is changed.
I mentioned your answer in GitHub issue. May be you could leave some suggestion there, if you got some spare minutes?
I mean, your opinion would be much more helpful :)
Guillaume Martres
@smarter
I know nothing about gradle or how they interact with zinc, so I can't say much
if you're invoking gradle multiple times from the command line, are you sure that it's reusing the same JVM instance ? Maybe something changed in how gradle handles JVMs
a cold JVM can easily be twice as slow at compiling scala than a hot JVM
otherwise, I guess you'll need to get a profiler and start measuring
Alex Shubert
@lure
Both builds use cold jvm. It's cloud CI agents. Anyway, I am going to measure module by module. Who knows, may be it is something else inside my own project. I just was surprised by the phrase about tradeoff.
Guillaume Martres
@smarter
yeah, try to reproduce it locally first
maybe the new gradle uses more memory but your JVM still has the same maximum heap, so you're spending more time doing GC, who knows
eugene yokota
@eed3si9n
the tradeoff likely comes from class-based name hashing introduced in Zinc 1 which does create more stuff to persist. 2x does seem like some pathological case. maybe you need to give more memory?