Status Update
Comments
vi...@google.com <vi...@google.com>
je...@google.com <je...@google.com>
je...@google.com <je...@google.com> #3
I was finally able to repro, and what happens is that if you finish the Gemini onboarding flow, you'll get the Ask Gemini
text, but you have to restart the IDE to get the link to show up.
We need to fix that, or tweak the text before rebooting to either hide it or say something like ("reboot to enable Ask Gemini feature").
nc...@meta.com <nc...@meta.com> #4
Background:
This feature is implemented by chaining together 2 separate pieces:
-
com/android/tools/idea/logcat/messages/MessageFormatter.kt
adds the text "Ask Gemini". For every message from logcat, we check if it is an exception, and if so, add the text "Ask Gemini". -
A separate
EditorHyperlinkDetector
is responsible for "linkifying" the text. But this detector, only adds the studio bot specific logic if studio bot was enabled during initialization.
Due to the inconsistency in how (1) and (2) detect whether studio bot is available, we run into this situation when:
- On IDE start up, logcat is initializated. Studio Bot is not enabled at this time, so the hyper link detector doesn't add the studio bot specific logic.
- User on boards to Studio bot. After this, the MessageFormatter will start adding the text "Ask Gemini" for exceptions.
I think we should do a simple fix right away: We should always add the StudioBotFilter
to the hyper link detector. This can then check for whether studio bot is enabled internally, which should make (1) and (2) consistent in how they check for studio bot being active.
je...@google.com <je...@google.com> #5
Can you also add an option to disable the feature, even if I logged in to Gemini?
We often share logcat output in bugreports and such. This is noise we can do without.
[Deleted User] <[Deleted User]> #6
We often share logcat output in bugreports and such. This is noise we can do without.
Just want to confirm: you are copying logcat output from the logcat pane in Studio in your bugreports? (These aren't added to the real adb logcat
output, they are only shown inside Studio, so they won't be in any automated bug report systems)
le...@gmail.com <le...@gmail.com> #7
This is a common practice at several teams I worked in, so it's not unique to my current workplace.
ca...@gmail.com <ca...@gmail.com> #8
Sounds good. Adding a setting is tracked in
je...@google.com <je...@google.com> #9
nc...@meta.com <nc...@meta.com> #10
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 Ladybug Feature Drop | 2024.2.2 Canary 1
- Android Gradle Plugin 8.8.0-alpha01
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!
ga...@google.com <ga...@google.com>
xa...@google.com <xa...@google.com> #11
We pushed RC2 without the fix because we needed to release it to validate some other more pressing issues.
The fix is pending. It's quite a large fix and we haven't fully decided whether to put it in 7.4.0 or wait to have it in 7.4.1. I believe it is in 8.0 Canary 10, so if you could try it there, that would be helpful. Ultimately though we have to also be sure if does not have other unforeseen side effects, and this may take a bit longer than we have time for (EE/7.4 are quite late and holding other releases so it's a bit tricky to handle this)
ga...@google.com <ga...@google.com> #12
The fix landed in AGP 8.0.0-alpha10
, as mentioned in #11. Updating ./gradlew :example:assembleDebug
successfully.
Once it lands in AGP 7.4 we'll update this bug with the exact version info. Thanks!
pe...@gmail.com <pe...@gmail.com> #13
Can't fully confirm, because it fails with another error now, because of React Native which doesn't work with it yet.
We really need this in 7.4. I don't really understand why it "missed the boat", since then 2 new RCs came out and it means many people won't be able to use 7.4 when it's release.
It's a fix for a showstopper regression. Should be tested in the wild asap.
ro...@sentry.io <ro...@sentry.io> #14
Did this fix make it to the 7.4.0 stable release?
nc...@meta.com <nc...@meta.com> #15
According to
On the React Native side of things, we had to un-bump AGP to 7.3 as this was a major blocker for us. We also have to tell users to don't update Android Studio as Electric Eel was released today and it will prompt users to update => Resulting in broken builds.
pe...@gmail.com <pe...@gmail.com> #16
ha...@gmail.com <ha...@gmail.com> #17
Everything is jammed till then
je...@google.com <je...@google.com> #18
So far, it is scheduled for Feb 1st which looks likely at this point.
[Deleted User] <[Deleted User]> #19
The error does not seem to be fixed in 7.4.1. Can you confirm this is indeed released?
nc...@meta.com <nc...@meta.com> #20
Actually the issue looks like fixed on our end. We're bumping React Native to use AGP 7.4.1 and the addGeneratedSourceDirectory
API here:
Thanks for working on this folks even if it took longer to land than expected
je...@google.com <je...@google.com> #21
#19, I am sorry, there was a misunderstanding with the release team. they had an even more urgent fix to push which was 7.4.1 and the normal dot release that contains this fix got pushed to 7.4.2 which is looking to be couple of weeks away (still working the date with them). So sorry for the wrong info, It surprised me too.
nc...@meta.com <nc...@meta.com> #22
That's confusing as the issue seems effectively fixed in 7.4.1, also in the reproducer I linked:
pe...@gmail.com <pe...@gmail.com> #23
For a workmate it also failed on his first try. We might see tomorrow.
Not sure yet, but it might be flaky now, depending on some cache or other internals.
I'll report back if I know more but it's good to know about the new target of 7.4.2. Really sad this kind of thing can't make it into a .0 release as prio 0 bug, especially since it *was* ready.
Maybe you guys can take this as a feedback and improve something, feels like it just hit a process / politics wall.
ar...@gmail.com <ar...@gmail.com> #24
je...@google.com <je...@google.com> #25
this might be due to a third party plugin. do you have a small project to reproduce the issue ?
can you provide a stack trace ?
Description
> Android Public Tracker > App Development > Android Studio > Gradle > Android Gradle Plugin
but I don't have permission to do so.
## Expected Result
AGP Variant API works well between 7.4.0-beta05 and 7.4.0-rc01
## Observed Results
Bumping AGP from 7.4.0-beta05 to 7.4.0-rc01 results in a build failure with the following message:
```
> Task :example:mapDebugSourceSetPaths FAILED
FAILURE: Build failed with an exception.
* What went wrong:
Execution failed for task ':example:mapDebugSourceSetPaths'.
> Querying the mapped value of map(provider(java.util.Set)) before task ':example:reproducerdebugTask' has completed is not supported
```
It seems like AGP is querying the value of the `@OutputDirectory` of the task I created `example:reproducerdebugTask`.
## Reproducer
You can find a small reproducer here:
Specifically this commit shows a working setup with AGP 7.4 beta5
-
Bumping to RC01 causes the failure:
-
I've also configured the CI so you have a clean environment where you can easily reproduce.
The command to reproduce upon cloning is `./gradlew :example:assembleDebug`.
## Further Context
This issue was noticed upstream on React Native here:
as we're using a similar pattern inside the React Native Gradle Plugin.