Status Update
Comments
hu...@google.com <hu...@google.com> #2
I can reproduce this (thanks for the repro project!)
It looks like the problem is that the desugared api list from r8 contains this entry:
java/util/Collection#removeIf(Ljava/util/function/Predicate;)Z
but the bytecode here doesn't match -- it's java/util/ArrayList. Collection isn't a directly implemented interface or a direct super class, it's an interface on the super super class. The most efficient thing runtime wise would be for the signature list to inline this method on all implemented subclasses. But I should probably at least for now go and make the desugared API lookup do something similar to what it does for API lookup -- search through all super classes and interfaces as well. This isn't a new problem, so I'm very surprised this hasn't come up before (or it has, and I've forgotten).
ku...@gmail.com <ku...@gmail.com> #3
(I have a pending CL that was working to improve the handling of fields now that r8 handles desugaring fields, I'll try to dust that off and combine the fix in there.)
ga...@google.com <ga...@google.com> #4
I went to implement this, and hooked up inheritance search when analyzing the source file containing the call.
However, lint also handles the case where the library being analyzed is not using core library desugaring (for example, it may be a plain Java library). But when that library is consumed in a downstream app module, where library desugaring is turned on, lint then processes the partial results from the library and filters each violation through the desugaring allowlist.
At this point, it's tricky to do the inheritance search -- this happens when we no longer have a compilation environment and can do class inheritance lookups. So there are three possible solutions.
First, we pay the cost up front -- even when you're not using core library desugaring, we record whether the method is potentially library desugared if turned on. (This is also tricky because at this point we don't know which exact desugaring library version is used, which determines the exact list of APIs).
Or, more expensively, for every API violation of this type we store all the potential super class and interface names for each result...
Or, we handle this in the code which generates the desugaring API list, inlining all subclasses affected. This could be quite a long list, but on the other hand this list is really only intended to be machine readable.
an...@gmail.com <an...@gmail.com> #5
This is partially fixed now; it's fully fixed for the sample project, but in the scenario I described in comment 4, it works if you also configure library desugaring to be on in the library.
au...@google.com <au...@google.com> #6
likely related:
Objects.requireNonNullElse
Objects.requireNonNullElseGet
are now also showing this same false positive warning.
Android Studio Koala | 2023.3.2 Canary 2 gradle 8.6 plugin: 8.3.1
an...@gmail.com <an...@gmail.com> #7
This seems unrelated. Can you file a new bug?
Specifically, for Objects.requireNonNullElse
I am not seeing a lint API error, but I am seeing a warning from IntelliJ's built-in Java inspections. These should probably not show up in Android modules. I swear we've fixed that before but I'm assuming something changed to break that.
For Objects.requireNonNullElseGet
I do see a lint API warning for that. But, that's not a recent change, right? I looked at R8's desugaring list and that method is not there. So that method does crash at runtime, right?
au...@google.com <au...@google.com> #8
Can you file a new bug?
fair enough - will do.
I swear we've fixed that before but I'm assuming something changed to break that.
I'm not 100% sure whether it first occurred when I moved to 8.3.1, and/or to Koala Canary 2
So that method does crash at runtime, right?
funnily enough no, even when I run on Android 8.0 (in the emulator); In fact I seem to have been using that method for 2 years now and the first of those 2 years I ran it personally on a physical device (Samsung A5 2017 with Android 8.0 / other users on?) ... and yes, I am flabbergasted as well :)
I'll try and dig a little regarding the requireNonNullElseGet first as indeed I realise it should crash, and then will file a new bug (as requireNonNullElse should work regardless)
I'll add a reference here for your enjoyment :)
an...@gmail.com <an...@gmail.com> #9
Actually it looks like D8 is really backporting it; I wrote this test class:
public class JavaTest {
public void test(String s) {
Objects.requireNonNullElse(s, "test");
Objects.requireNonNullElseGet(s, new Supplier<String>() {
@Override
public String get() {
return null;
}
});
}
}
Built the APK and opened the APK analyzer, drilled to this method and invoked Show Bytecode and here's what I get:
# virtual methods
.method public test(Ljava/lang/String;)V
.registers 3
.param p1, "s" # Ljava/lang/String;
.line 8
const-string v0, "test"
invoke-static {p1, v0}, Lcom/example/myapplication/JavaTest$$ExternalSyntheticBackport0;->m(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
.line 9
new-instance v0, Lcom/example/myapplication/JavaTest$1;
invoke-direct {v0, p0}, Lcom/example/myapplication/JavaTest$1;-><init>(Lcom/example/myapplication/JavaTest;)V
invoke-static {p1, v0}, Lcom/example/myapplication/JavaTest$$ExternalSyntheticBackport1;->m(Ljava/lang/Object;Ljava/util/function/Supplier;)Ljava/lang/Object;
.line 15
return-void
.end method
So both methods are desugared.
Clement/Søren, I don't see this in the backport list:
$ java -cp $ANDROID_HOME/cmdline-tools/latest/lib/r8.jar com.android.tools.r8.BackportedMethodList --min-api 15 | grep requireNonNullElse
java/util/Objects#requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
(requireNonNullElse
is there, requireNonNullElseGet
is not.)
hm...@google.com <hm...@google.com> #10
aha - so requireNonNullElseGet is not magical after all :)
Please let me know if I should still file a new bug.
tx
sf...@netflix.com <sf...@netflix.com> #11
The method Objects.requireNonNullElseGet
is a bit special in terms of backporting, as it takes an argument of type Supplier
, which was added in API level 24. For pure backporting (no desugared library) this method can only be backported from API level 24:
$ java -cp $ANDROID_HOME/cmdline-tools/latest/lib/r8.jar com.android.tools.r8.BackportedMethodList --min-api 24 | grep requireNonNullElse
java/util/Objects#requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
java/util/Objects#requireNonNullElseGet(Ljava/lang/Object;Ljava/util/function/Supplier;)Ljava/lang/Object;
Setting minSdk
below 24 (without using desugared library) this is the DEX for the test code.
.method public test(Ljava/lang/String;)V
.locals 1
.param p1, "s" # Ljava/lang/String;
.line 8
const-string v0, "test"
invoke-static {p1, v0}, Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest$$ExternalSyntheticBackport0;->m(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
.line 9
new-instance v0, Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest$1;
invoke-direct {v0, p0}, Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest$1;-><init>(Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest;)V
invoke-static {p1, v0}, Ljava/util/Objects;->requireNonNullElseGet(Ljava/lang/Object;Ljava/util/function/Supplier;)Ljava/lang/Object;
.line 15
return-void
.end method
with only Objects.requireNonNullElse
getting backported.
With desugared library enabled this change, due to two things
- desugared library adds support for
Supplier
Objects
is no longer backported, but implemented by desugared library
The "tool" com.android.tools.r8.ir.desugar.desugaredlibrary.lint.DesugaredMethodsList
is used for desugared library:
java -cp $ANDROID_HOME/cmdline-tools/latest/lib/r8.jar com.android.tools.r8.ir.desugar.desugaredlibrary.lint.DesugaredMethodsList --min-api 15 --desugared-lib META-INF/desugar/d8/desugar.json --lib $ANDROID_HOME/platforms/android-34/android.jar --desugared-lib-jar desugar_jdk_libs-2.0.4.jar | grep Objects
java/util/Objects
(META-INF/desugar/d8/desugar.json
is from desugar_jdk_libs-2.0.4.jar
is
The line java/util/Objects
indicate that all methods of Objects
are supported.
With desugared library (for any minSdk
) this is the DEX for the test code:
# virtual methods
.method public test(Ljava/lang/String;)V
.locals 1
.param p1, "s" # Ljava/lang/String;
.line 8
const-string v0, "test"
invoke-static {p1, v0}, Lj$/util/Objects;->requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
.line 9
new-instance v0, Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest$1;
invoke-direct {v0, p0}, Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest$1;-><init>(Ldk/gjesse/jdk11desugaredlibrary/ObjectsTest;)V
invoke-static {p1, v0}, Lj$/util/Objects;->requireNonNullElseGet(Ljava/lang/Object;Ljava/util/function/Supplier;)Ljava/lang/Object;
.line 15
return-void
.end method
hm...@google.com <hm...@google.com> #12
Thanks for that explanation Søren.
So, when I tried this with the latest Koala canary, it looks like things are working correctly. For the following code:
Objects.requireNonNullElse
Objects.requireNonNullElseGet
Both AGP and Studio correctly will not flag the first call.
The second call is flagged as an API error. But if I turn on core library desugaring, then that warning also goes away, both in AGP and in Studio.
There is still the issue of the builtin Java 9 APIs inspection in IntelliJ -- again, not a lint bug, but it is a Studio bug. Original submitter, can you file that one? And can you confirm (if you're not seeing warnings for these) that you are in fact using library desugaring and a recent Studio?
sf...@netflix.com <sf...@netflix.com> #13
I've filed
au...@google.com <au...@google.com> #14
thanks for filing the bug!
using library desugaring and a recent Studio?
yes I am, I have desugaring in all modules + using Koala Canary 2 now.
I will have to switch back to Jelly fish anyhow due to Koala being (to put it mildly) a runaway memory and cpu lover... which makes it nearly unusable (but that's another issue) so I might not be able to follow up on the new bug in the short term.
sf...@netflix.com <sf...@netflix.com> #15
On studio-main I can still repro the following false positive from
- Create a new project
- Enable desugaring, run Gradle sync
- Add a usage of
java.util.Objects#requireNonNullElse
Lint warning:
Call requires API level 30 (current min is 26): java.util.Objects#requireNonNullElse
au...@google.com <au...@google.com> #16
Just tested this with Koala.1 Beta 1, and something seems to not be right with lint (see results below). Not exactly sure where the integration is not working.
Using the test from git clone sso://user/sgjesse/B327670482
minApi | desugared library | requireNonNullElse | equireNonNullElseGet |
---|---|---|---|
21 | No | No lint message | Call requires API level 30 (and Call requires API level 24 on Supplier ) |
24 | No | No lint message | Call requires API level 30 |
21 | Yes | Call requires API level 30 | Call requires API level 30 |
The expected messages should be:
minApi | desugared library | requireNonNullElse | equireNonNullElseGet |
---|---|---|---|
21 | No | No lint message | Call requires API level 30 (and Call requires API level 24 on Supplier ) |
24 | No | No lint message | No lint message |
21 | Yes | No lint message | No lint message |
Looking at the DEX after compiling the code it is desugared as expected:
minApi | desugared library | requireNonNullElse | equireNonNullElseGet |
---|---|---|---|
21 | No | Desugared (backport) | Call to java.util.Objects through an D8 outline |
24 | No | Desugared (backport) | Desugared (backport) |
21 | Yes | Desugared (to target j$.util.Objects ) |
Desugared (to target j$.util.Objects ) |
Running the command line version of the D8 lint info generation tool the output looks correct.
As far as I can see the BackportedMethodList
D8 tool generated the correct desugared methods:
minApi
21 no library desugaring:
java -cp build/libs/r8.jar com.android.tools.r8.BackportedMethodList --min-api 21 --lib third_party/android_jar/lib-v34/android.jar | grep Objects
java/util/Objects#checkFromIndexSize(III)I
java/util/Objects#checkFromIndexSize(JJJ)J
java/util/Objects#checkFromToIndex(III)I
java/util/Objects#checkFromToIndex(JJJ)J
java/util/Objects#checkIndex(II)I
java/util/Objects#checkIndex(JJ)J
java/util/Objects#isNull(Ljava/lang/Object;)Z
java/util/Objects#nonNull(Ljava/lang/Object;)Z
java/util/Objects#requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
requireNonNullElse
is desugared.
minApi
24 no library desugaring:
java -cp build/libs/r8.jar com.android.tools.r8.BackportedMethodList --min-api 24 --lib third_party/android_jar/lib-v34/android.jar | grep Objects
java/util/Objects#checkFromIndexSize(III)I
java/util/Objects#checkFromIndexSize(JJJ)J
java/util/Objects#checkFromToIndex(III)I
java/util/Objects#checkFromToIndex(JJJ)J
java/util/Objects#checkIndex(II)I
java/util/Objects#checkIndex(JJ)J
java/util/Objects#requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
java/util/Objects#requireNonNullElseGet(Ljava/lang/Object;Ljava/util/function/Supplier;)Ljava/lang/Object;
requireNonNullElse
and requireNonNullElseGet
are desugared.
minApi
21 with library desugaring:
java -cp build/libs/r8.jar com.android.tools.r8.BackportedMethodList --min-api 21 --lib third_party/android_jar/lib-v34/android.jar --desugared-lib src/library_desugar/jdk11/desugar_jdk_libs.json | grep Objects
java/util/Objects#checkFromIndexSize(JJJ)J
java/util/Objects#checkFromToIndex(JJJ)J
java/util/Objects#checkIndex(JJ)J```
Neither requireNonNullElse
nor requireNonNullElseGet
is mentioned, but META-INF/desugar/d8/lint/compile_api_level_30/desugared_apis_30_1.txt
from the desugared library configuration artifact contains:
java/util/Objects
So all of Objects
is desugared.
We need to track down where the information is lost.
sf...@netflix.com <sf...@netflix.com> #17
Shouldn't this be a P1 and Koala.1 RC blocker?
au...@google.com <au...@google.com> #18
not as far as I'm concerned - this is a warning only which I can just ignore. The app compiles and works perfectly fine.
Methinks you guys have more important P1's to fix. If you need inspiration, then look at #127100532 (not related to this one at all, I'm just being ... now)
sf...@netflix.com <sf...@netflix.com> #19
I think I found the problem.
Here's the desugaring files used in the IDE:
modelArtifact?.desugaredMethodsFiles = {ArrayList@83951} size = 2
0 = {File@84956} "/Users/tnorbye/.gradle/caches/transforms-4/5126fdfcbb5a995f09b8be45eb610aef/transformed/desugar_jdk_libs_configuration_nio-2.0.4-desugar-lint.txt"
1 = {File@84957} "/Users/tnorbye/.gradle/caches/transforms-4/fc5ceb7b5de2ee8487ba96f4a79c6608/transformed/D8BackportedDesugaredMethods.txt"
What we do with these files is to "merge" them. This was just reading all the signatures from both files and sorting them. When lint is checking methods it just does a binary search.
But in this case, we're getting these methods from the d8 backport list:
java/util/Objects#checkFromIndexSize(JJJ)J
java/util/Objects#checkFromToIndex(JJJ)J
java/util/Objects#checkIndex(JJ)J
And we're getting this method from the library desugaring list:
java/util/Objects
This means that we end up with this in the signature list:
java/util/Objects
java/util/Objects#checkFromIndexSize(JJJ)J
java/util/Objects#checkFromToIndex(JJJ)J
java/util/Objects#checkIndex(JJ)J
This is wrong; the first line implies that all the other three are included. But in a binary search, if our midpoint hits one of the specific methods, we'll conclude that the match must be later in the list, so we'll falsely decide the method isn't there.
I need to fix the merging of signature files to not just concatenate but to drop specific methods and fields if the other file lists the whole class.
I don't think this is a recent regression, but I think this is showing up now through a combination of the desugaring files including newly fully backported classes, and there's luck involved based on how binary search proceeds.
au...@google.com <au...@google.com> #20
So I see two bugs listed in your table in
For the second one, I see that lint is passed this single database file: /Users/tnorbye/.gradle/caches/transforms-4/6cf6a445b18927988f27d9e2411efbbe/transformed/D8BackportedDesugaredMethods.txt
And that file does not contain requireNonNullElseGet
:
$ grep Objects /Users/tnorbye/.gradle/caches/transforms-4/6cf6a445b18927988f27d9e2411efbbe/transformed/D8BackportedDesugaredMethods.txt | grep requireNon
java/util/Objects#requireNonNull(Ljava/lang/Object;)Ljava/lang/Object;
java/util/Objects#requireNonNull(Ljava/lang/Object;Ljava/lang/String;)Ljava/lang/Object;
java/util/Objects#requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
Is this a problem in AGP/sync? cc: Scott
ju...@outsystems.com <ju...@outsystems.com> #21
Found the issue in AGP for the second bug. When calling D8DesugaredMethodsGenerator.generate
the minSdk
is not passed, so it defaults to 1. Therefore changing from 21 to 24 does not make a difference (except for the red squiggles under Supplier
goes away). When building the BackportedMethodListCommand
an additional call to setMinApiLevel
on the builder with the actual minSdk
is missing
Running the command line version of the tool without --min-api
gives exactly the list in
java -cp build/libs/r8.jar com.android.tools.r8.BackportedMethodList --lib third_party/android_jar/lib-v34/android.jar | grep Objects | grep requireNon
java/util/Objects#requireNonNull(Ljava/lang/Object;)Ljava/lang/Object;
java/util/Objects#requireNonNull(Ljava/lang/Object;Ljava/lang/String;)Ljava/lang/Object;
java/util/Objects#requireNonNullElse(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;
an...@gmail.com <an...@gmail.com> #24
(2) With core library desugaring off, and minSdkVersion 24, lint flags requireNonNullElseGet
Fixed with Ib00094d61dbc26ff49928855ebe8dcdedf5ecf48
Close the ticket based on
js...@google.com <js...@google.com> #25
Thank you for your patience while our engineering team worked to resolve this issue. A fix for this issue is now available in:
- Android Studio Koala Feature Drop | 2024.1.2 Canary 6
- Android Gradle Plugin 8.6.0-alpha06
We encourage you to try the latest update.
If you notice further issues or have questions, please file a new bug report.
Thank you for taking the time to submit feedback — we really appreciate it!
ho...@netflix.com <ho...@netflix.com> #26
glad to confirm this is now working fine :)
Even with Plugin 8.5.0
au...@google.com <au...@google.com> #27
artifacts compatible with ancient versions of AGP other than stop
publishing compose desktop artifacts.
On Thu, Apr 4, 2024, 5:01 PM hokanla <buganizer-system@google.com> wrote:
an...@gmail.com <an...@gmail.com> #28
if I was using androidx directly by myself I would not be complaining too much as there are some ways around it. but since I provide a library that customers use, and some of those customers are React-native, and Ionic (similar to Unity mentioned above) i will have to permanently add this instruction and they will have to implement an extra set of code in their app just to make my library work with their project. It is probably unexpected that such dependency would come from a medium version bump if that makes sense.
Again I'm not sure if there is a better way around it, probably there isnt aside from getting AGP up to that. its just not the reality for most of production projects which tends to lag behind, particularly if we talk about xplatform projects, since people might not even know what gradle is
hm...@google.com <hm...@google.com> #29
As AGP we can't do much about this fact that people can't upgrade to the latest version. Essentially the issue stems from a discrepancy in the way KGP publishes multiplatform libraries and the way AGP consumes them.
This problem is now fixed and in AGP 8.4 RC2 (and 8.5 canary 4).
From KGP side the problem is also fixed - please refer to
Additionally, if you are a library developer and want to target android only you could consider not depending on androidx.compose.runtime:runtime (which exposes -android and -desktop variants) but directly depend on the androidx.compose.runtime:runtime-android (
au...@google.com <au...@google.com> #30
Additionally, if you are a library developer and want to target android only you could consider not depending on androidx.compose.runtime:runtime (which exposes -android and -desktop variants) but directly depend on the androidx.compose.runtime:runtime-android (
https://maven.google.com/web/index.html?#androidx.compose.runtime:runtime-android ) artifact - I think this is should work - feel free to correct me if not.
it does not really work that way as each -android variant will pull in root non-android ones.
an...@gmail.com <an...@gmail.com> #31
so...@google.com <so...@google.com> #32
A postmortem task was automatically filed as a child item to this bug.
an...@google.com <an...@google.com> #33
Thank you for your patience while our engineering team worked to resolve this issue. A fix for this issue is now available in:
- Android Studio Jellyfish | 2023.3.1
- Android Gradle Plugin 8.4.0
We encourage you to try the latest update.
If you notice further issues or have questions, please file a new bug report.
Thank you for taking the time to submit feedback — we really appreciate it!
[Deleted User] <[Deleted User]> #34
Help
[Deleted User] <[Deleted User]> #35
Help
Description
When upgrading Compose from 1.5 -> 1.6, the Application stops working and immediately crashes with:
```
java.lang.NoClassDefFoundError: Failed resolution of: Landroidx/compose/ui/platform/ComposeView;
```
Issue repro:
Build Scan:
The build scan shows that after upgrading to Compose 1.6, the app starts being packaged with `desktop` compose artifacts rather than `android` artifacts. This seems to happen because the application module does not apply the kotlin gradle plugin. Once I apply the kotlin plugin to the app module (even though that would be unnecessary) the issue goes away.
For projects that have a combination of Kotlin/Compose and Java modules, another issue is that the build resolves compose-desktop artifacts during the build process for any java module that happens to have a transient dependency on compose.
STEPS TO REPRODUCE:
1. Create a project with library module (compose + kotlin) and app module (pure Java)
2. Upgrade compose runtime dependencies to 1.6+
3. Observe crash on launch, android compose artifacts not included in APK
------------------
Studio Build: Jellyfish
Version of Gradle Plugin: 8.2.2
Version of Gradle: 8.6
Version of Java: 17
OS: macOS