Interesting, with gradle 4.10.2 command `./gradlew...
# gradle
c
Interesting, with gradle 4.10.2 command
./gradlew clean classes --parallel
takes about 1.5 minutes on my machine, with gradle 5.0, same command, same project, same machine - about 12 minutes... I thought gradle 5.0 was supposed to be faster 😄 Seems like most time is spent on dependencies resolution. Are there any known problems with that? Especially when using Spring dependency-management plugin?
s
Just curious, should we need to use
Spring dependency-management
plugin in 5.0? Can’t
implementation(enforcedPlatform())
work for springboot apps?
o
could be the lower memory limit is incurring more GC cycles, making it take much longer, but that seems like it would be too minor to cause a 10 minute addition
a
Might be caused by connection/transfer latency
c
I've read that
implementation(enforcedPlatform())
is not doing same things as Spring's plugin, but I'll try it at some point. As for latency, I don't see how it would, it's same network and all deps are already cached.
a
Does it happen every time or first time after you've updated to Gradle 5.0? I might be wrong, but Gradle might verify all cached deps with their remote versions on Gradle updates For context: we had a lot of problems with JCenter and other Bintray repos this year, often connection & transfer timeouts and increased latency that resulted in very long builds (since then we host own artifactory)
c
every time, it is very consistent unfortunately.
a
c
Seems to be the case, will check tomorrow. Thank you.
The workaround specified in the issue comments solved the problem for command line execution, unfortunately no effect on IntelliJ refresh. I tried both configuring it through gradle.properties:
systemProp.org.gradle.internal.repository.max.tentatives=1
and through "Gradle VM options" in IntelliJ settings:
-Dorg.gradle.internal.repository.max.tentatives=1
o
those are JVM options
you should try
-D<key>=<value>
like on CLI
c
Yes, I did it with
-D
, only missed it when copy-pasting here. Doesn't work.