Status Update
Comments
sg...@google.com <sg...@google.com>
sg...@google.com <sg...@google.com> #2
Gradle issue asking for it to be made easier to avoid making an unused modification to this list:
ph...@gmail.com <ph...@gmail.com> #3
I think the issue here is that jvmArgs is null, which causes addAll()
to not run.
IMO we should be adding to jvmArgs and not replacing it entirely, in case something else has previously set jvmArgs. WYT @Ivan?
I could change it to something like
forkOptions.jvmArgs?.addAll(executionOptions.get().jvmArgs) ?: forkOptions.jvmArgs(executionOptions.get().jvmArgs)
ra...@gmail.com <ra...@gmail.com> #4
Yeah, the nullness of jvmArgs
was the first possibility I checked too, although even with "!!" it still had no effect, see
I uploaded a change that I observed to fix it for me and added you as a reviewer
ap...@google.com <ap...@google.com> #5
Ah, ok, that makes sense.
ap...@google.com <ap...@google.com> #6
Thanks for the change, Jeff!
Should we cherry-pick it into 7.4? I think AGP 7.4 has that but the settings plugin is not officially supported yet.
ap...@google.com <ap...@google.com> #8
ap...@google.com <ap...@google.com> #9
Branch: 8.0
commit dbb34e52278af4bf44beba8c84abfd9e2d2a753a
Author: Søren Gjesse <sgjesse@google.com>
Date: Wed May 10 09:24:24 2023
Use rematerializable candidate with short live range, if no alternative
Bug:
Change-Id: I40782eeede946678c21544454d675329634d21be
M src/main/java/com/android/tools/r8/ir/regalloc/LinearScanRegisterAllocator.java
ap...@google.com <ap...@google.com> #10
Branch: 8.1
commit e7fc08aa7e632f3b6123d94bf2d8bdd482131194
Author: Søren Gjesse <sgjesse@google.com>
Date: Wed May 10 09:23:30 2023
Use rematerializable candidate with short live range, if no alternative
Bug:
Change-Id: I40782eeede946678c21544454d675329634d21be
M src/main/java/com/android/tools/r8/ir/regalloc/LinearScanRegisterAllocator.java
Description
I started seeing
minifyReleaseWithR8
task failure of my Compose project. It fails with some random project library jar, e.g.If I exclude this library from proguard, it'll fail with some next library - it can be both Compose one or any other, looks like message is unrelated to the problem.
By commenting code in the project, I've found out that I can fix the build by replacing one of
@StringRes param: Int
in composable function withparam: String
. But I still not sure if that's the root case.When trying to reproduce it in a sample project, I was able to reproduce it with even simpler code, that doesn't include int parameter. If I switch from compose-material to compose-material3, it builds fine.
MRE with dependencies is attached.