Skip to content

fix(android): Nav3 support for ViewPager2 - #6398

Merged
0xadam-brown merged 3 commits into
mainfrom
0xadam-brown/collection/nav3-deferred-transactions
Nov 4, 2026
Merged

0xadam-brown merged 3 commits into
mainfrom
0xadam-brown/collection/nav3-deferred-transactions

Conversation

@0xadam-brown

@0xadam-brown 0xadam-brown commented Oct 27, 2026 •

Copy link
Copy Markdown
Member

Fyi, I wrote the PR description below by hand. It's beefy, but it has some useful background if you want it. (No obligation. It's helpful for my future reference either way.)

PR Stack (Stage Nav3 Transactions)


📜 Description

Makes the multi-Activity and multi-Fragment support introduced by the #6301 PR stack play nicely with ViewPager2.

If you want the tl;dr, start at the "Solution" section below.

💡 Motivation and Context

Nothing is free: Supporting multi-Activity and multi-Fragment architectures

The #6301 PR stack implemented multi-Activity and multi-Fragment support for our key Nav3 integration API, SentryNavEffect, and the BackStackObserver class that supports it. (BackStackObserver consumes updates to host app's back stack and translates them into nav data + nav transactions.)

Explicit multi-Activity/-Fragment support was needed because the initial implementation of SentryNavEffect assumed there would be one and only one effect active at a given time.** That's true for single-Activity architectures, but not for multi-Activity or multi-Fragment ones in which each Activity or Fragment has its own nav subgraph and therefore its own SentryNavEffect. (E.g., if Activity A and Activity B each have their own SentryNavEffect, those effects will be briefly active at the same time when the user navigates from Activity A to B, or vice versa.) Absent the changes in #6301, multi-Activity / - Fragment architectures could generate stale or misleading nav data.

** We assumed that only one SentryNavEffect would be active at a time in the interest of releasing core Nav3 support more quickly, and we documented the limitation instead. The Sentry Android SDK has a long-standing policy of sharing a single IScopes instance throughout the SDK with a single underlying IScope, which is used to accumulate Sentry data and hold the currently active transaction. Assuming SentryNavEffect had exclusive access to that IScope simplified our initial implementation considerably. (Changing that policy was well outside the bounds of our Nav3 work, of course.) #6301 was all about breaking that assumption.

Almost there: The original approach to multi-Activity / -Fragment support

The trick of #6301 was to ensure that only one SentryNavEffect at a time could write to the shared IScope instance. It did so by replacing the IScopes object used by BackStackObserver with a new NavScopes facade that lets us swizzle the destination of nav data updates made by BackStackObserver. Updates to the underlying IScope were published directly whenever a given SentryNavEffect <> BackStackObserver <> NavScopes tuple had an exclusive lease; otherwise, they were shunted to a staging area.

Staged updates were published later if the SentryNavEffect tuple eventually acquired the lease. The lease could be acquired whenever both i) the back stack tracked by the SentryNavEffect was initialized or updated, and ii) the lease wasn't currently held by another SentryNavEffect tuple in a resumed / visible state. (If one or both conditions failed, the requesting SentryNavEffect would be denied the lease and would continue to write to its staging area. Note, as well, that in most cases satisfying both conditions implies that the requesting SentryNavEffect is itself resumed / visible. That'll matter for the discussion below.)

Combined, those^^ updates let #6301 separate navigation timing (in the form of host back stack updates) from publication of nav data. And that, in turn, saved us from having to drop nav data or let it become stale as the user switched between Activities or Fragments with their own nav subgraphs. Hooray.

The next hurdle: ViewPager2

...but it turns out we'd neglected an important use case, viz., ViewPager2.

The approach in #6301 was smart about staging nav data (ie, screen name + nav context), but it didn't do the same for nav transactions. A new nav transaction could be started only if the lease was owned at the time of navigation, where "navigation" == initialization or modification of the nav back stack. Because the lease had to be owned, that meant that the Activity or Fragment hosting the SentryNavEffect requesting the lease had to be in a resumed state.

That assumption made sense for most use cases, because usually initialization of an Activity or Fragment's nav back stack coincides with the Activity or Fragment being visible and therefore being able to preempt the lease from the previous leaseholder. But the assumption didn't pan out for ViewPager2, because ViewPager2 preloads neighboring pages. That means it initializes the nav back stack for Fragments that aren't in a resumed state and therefore can't preempt the lease, causing them to drop their initial nav transaction and any spans created by the initial destination screen. Sigh.

More details about ViewPager2

Unlike most Activity and Fragment architectures, ViewPager2 creates neighboring pages before a user navigates to them (and therefore before they become visible). When ViewPager2 creates a page, it populates the page's SentryNavEffect with an initial nav back stack. (Note that each page gets its own nav back stack, and there's no nav back stack that tracks movements between the ViewPager2 pages.)
.
Pre-population of a page's nav back stack is first chance SentryNavEffect has to try to claim the lease and start a nav transaction. If the page is currently visible, the lease will be acquired, the nav transaction will be published, and all will be well. But if the page is pre-loaded before it's visible, the initial attempt to acquire the lease will fail and the nav transaction will be dropped (at least under the approach from #6301). The user's later swiping to the pre-loaded page won't modify that page's back stack, and so no additional request will be made to start a nav transaction. Oof.

The solution

The solution this PR stack implements is to create a mechanism similar to the data staging from #6301, but for transactions. Doing so lets us separate back stack initialization observed during preloading vs attempted lease acquisition when a ViewPager2 page becomes visible.

In particular, the new StagedNavTransaction class tracks requests to start nav transactions when a SentryNavEffect is created even if it doesn't currently have the lease. If the lease is later acquired and the transaction slot is available on the shared IScope, the staged transaction is started, which keeps ViewPager2 happy.

Before and after

 BEFORE #6398 
  ============ 
 
  Page A visible                  Page B preloaded offscreen 
  A holds nav lease               B observes initial destination 
          |                                   | 
          |                         Attempts nav transaction 
          |                                   | 
          |                         No lease -> request declined (page A is still visible, so page B can't preempt the lease)
          |                                   | 
          +---------- swipe to B ------------>| 
                                              | 
                                    B acquires nav lease 
                                    Publishes screen + back stack 
                                              | 
                                    Destination hasn't changed 
                                    No new transaction attempt 
                                              | 
                                    NO B NAV TRANSACTION 
 AFTER #6398 
  =========== 
 
  Page A visible                  Page B preloaded offscreen 
  A holds nav lease               B observes initial destination 
          |                                   | 
          |                         Requests nav transaction 
          |                                   | 
          |                         No lease -> request staged rather than dropped
          |                                   | 
          +---------- swipe to B ------------>| 
                                              | 
                                    B acquires nav lease 
                                    Publishes screen + back stack 
                                              | 
                                    Attempts deferred start to nav transaction
                                              | 
                                +-------------+-------------+ 
                                |                           | 
                          Scope available             Another active 
                                |                   transaction blocks it 
                                v                           | 
                       START B TRANSACTION                  v 
                                                     DISCARD REQUEST 
                                                     No later retry

Returning to a previously active page follows the same rule before and after this PR stack:

  B loses lease -> B's transaction is unbound; timers keep running 
          | 
          +---------- swipe back to B ----------+ 
                                                | 
                                    Did B navigate without 
                                    the lease while away? 
                                                | 
                           +--------------------+--------------------+ 
                           |                                         | 
                          NO                                        YES 
                           |                                         | 
                Existing B transaction?                 BEFORE: request declined 
                           |                            No new transaction on return 
                   +-------+-------+ 
                   |               |                    AFTER: latest request staged 
               Unfinished       Finished                Attempt start on return; 
                   |               |                    discard if blocked 
                   v               v 
                REBIND         NO TRANSACTION 
             existing one        STARTS 

A swipe alone creates no new navigation request. The stack lets a swipe fulfill a request recorded earlier, during preloading or navigation without the lease.

⚠️ Callouts

As shown in the final diagram above, this PR stack implements an approach that starts a nav transaction the first time the user swipes to a ViewPager2 page, but not upon swiping back between already visited pages. (Note: Previous transactions are kept in a idling state, so they can be resumed, but if they expire they won't be restarted.)**

That's contrary to what I had originally proposed here:

My plan (for now) is to start a txn on lease acquisition by replaying the page's last back stack through the observer, so that a page that acquires the lease w/o a txn will be able to start one in the usual way. It'll be timed from when the swipe settles, which for a preloaded page is when the work the user can see actually starts. Chime in if you think that's bonkers (now or when you see the PR). A lot of corner cases in this... :blob-sweat:

In the end I deviated b/c:

  1. I discovered that the current approach (ie, in this PR) matches what we do for ui.load transactions;
  2. it also matches what we do for Nav2;
  3. our goal for SentryNavEffect generally has been parity with Nav2;
  4. the current approach closes the most problematic behavioral gaps;
  5. the bulk of the work in this PR stack is refactoring that we'll want regardless of the policy we choose; and
  6. it's probably preferable to apply a single policy for all nav roots (Activity, Activity-hosted Fragment, or ViewPager2-hosted Fragment) – meaning even if we want Nav3 to diverge from Nav2 and ui.load transactions, we should consider whether we're okay diverging for all Activities and Fragments and not just in the case of ViewPager2.

If folks prefer, we could consider an across-the-board policy that starts a new transaction whenever the SentryNavEffect root changes (and not just when SentryNavEffect's back stack is initialized or updated). The details are tracked under #6454.

** This isn't as bad as it sounds! First, if the ViewPager destroys a previously visited page, then a new transaction will be started for it the next time the user swipes back to it. Second, usually each ViewPager2 page will be the root of a composable nav subgraph. So long as the user navigates inside that page, new transactions will be started. It's just when swiping the ViewPager between already created pages that no transaction is restarted, since the ViewPager2 is bringing existing Fragments back into resumed state without modifying their nav back stacks.

More details The BackStackObserver associated with each SentryNavEffect ensures that only nav back stack updates trigger requests to start nav transactions. Creating an Activity or Fragment (or ViewPager2 Fragment) produces a nav back stack "update" when the nav back stack is initialized; resuming a non-resumed Activity or Fragment does not. That leads to two issues:
  1. resuming existing Activities or Fragments, each with their own SentryNavEffect, does not trigger a nav change and therefore doesn't request a new nav transaction or publish new nav data to IScope;
  2. starting nav transactions or re-publishing nav data is something we have to do manually when an existing Activity or Fragment is resumed.

Both #6301 and this PR stack re-publish nav data when an existing Activity or Fragment is resumed so that IScope data doesn't reflect the stale outgoing Activity or Fragment's nav destination. Both also rebind the incoming Activity's or Fragment's transaction. If that transaction hasn't timed out, any spans created by the incoming Activity or Fragment will be parented under the resumed nav transaction.

The question the "across-the-board policy" mentioned above raises is whether we should create a new transaction instead of rebinding the incoming Activity or Fragment's existing transaction. The latter may have expired, whereas the former is guaranteed to be fresh. The downside of a new transaction is that it creates a discontinuous span context.

addresses: JAVA-785

💚 How did you test it?

See the stack PRs.

📝 Checklist

  • I added GH Issue ID & Linear ID
  • I added tests to verify the changes.
  • No new PII added or SDK only sends newly added PII if sendDefaultPII is enabled.
  • I updated the docs if needed.
  • I updated the wizard if needed.
  • Review from the native team if needed.
  • No breaking change or entry added to the changelog.
  • No breaking change for hybrid SDKs or communicated to hybrid SDKs.
  • Public API changes reviewed by another Mobile SDK team member or implemented according to the develop docs spec.

🔮 Next steps

#skip-changelog

  • sentry-docs: Update Sentry Android Navigation page to document ViewPager2 support and limitations.
  • If desired, fast-follow with an across-the-board policy that starts a new transaction whenever the SentryNavEffect root changes (and not just when SentryNavEffect's back stack is initialized or updated).

@sentry

sentry Bot commented Oct 27, 2026 •

Copy link
Copy Markdown

📲 Install Builds

Android

🔗 App Name App ID Version Configuration
SDK Size io.sentry.tests.size 8.62.0 (1) release

⚙️ sentry-android Build Distribution Settings

@0xadam-brown 0xadam-brown changed the title fix(android): Nav3 Deferred Transactions fix(android): Stage Nav3 Transactions Oct 27, 2026
@0xadam-brown 0xadam-brown added the sanity-check PR needs a lightweight review for obvious issues label Oct 27, 2026
@0xadam-brown
0xadam-brown marked this pull request as ready for review October 27, 2026 23:20
@0xadam-brown
0xadam-brown requested review from a team, markushi and runningcode and removed request for a team November 3, 2026 08:32
@0xadam-brown 0xadam-brown changed the title fix(android): Stage Nav3 Transactions fix(android): Nav3 support for ViewPager2 Nov 3, 2026
…ransactions (JAVA-785) (#6399)

Refactor .navigation3.lifecycle ahead of staging nav3 transactions when they're requested by a nav root without a lease. (Staging transactions is required in order to better support ViewPager2 pages and other scenarios that pre-build their back stacks before their nav roots are displayed.)

The refactor lets us separate three responsibilities:

- Lease coordination: NavLeaseCoordinator observes registered owners’ lifecycles and decides which SentryNavEffect may publish navigation state.

- State publication: Each effect’s NavScopes publishes and withdraws its navigation data and transaction together when ownership changes.

- Transaction management: Each effect gets a transaction handle that manages starting, finishing, binding, and restoring its transaction. The coordinator no longer retains transactions itself.

Behavior is unchanged save for one corner case: when a SentryNavEffect regains the lease, its unfinished transaction is now also re-bound if the scope holds a finished transaction. Before, it was only re-bound into an empty scope. The update matches the rule for starting a transaction, and in practice shouldn't occur, since a finished SentryTracer removes itself from the scope.

@cursor cursor Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Fix All in Cursor

❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.

Want reviews to match your repository better? Bugbot Learning can learn team-specific rules from PR activity. A team admin can enable Learning in the Cursor dashboard.

Reviewed by Cursor Bugbot for commit 0d30809. Configure here.

Make the multi-Activity and multi-Fragment support introduced by the #6301 PR stack play nicely with ViewPager2.

BACKGROUND

The approach in #6301 was smart about staging nav data (ie, screen name + nav context), but it didn't do the same for nav transactions. A new nav transaction could be started only if the lease was owned at the time of navigation, where "navigation" == initialization or modification of the nav back stack. Because the lease had to be owned, that meant that the Activity or Fragment hosting the SentryNavEffect requesting the lease had to be in a resumed state.

That^^ assumption made sense for most use cases, because usually initialization of an Activity or Fragment's nav back stack coincides with the Activity or Fragment being visible and therefore being able to preempt the lease from the previous leaseholder. But the assumption didn't pan out for ViewPager2, because ViewPager2 preloads neighboring pages. That means it initializes the nav back stack for Fragments that aren't in a resumed state and therefore can't preempt the lease, causing them to drop their initial nav transaction and any spans created by the initial destination screen. Sigh.

THE SOLUTION

The solution this commit implements is to create a mechanism similar to the data staging from #6301, but for transactions. Doing so lets us separate back stack initialization observed during preloading vs attempted lease acquisition when a ViewPager2 page becomes visible.

In particular, the new StagedNavTransaction class tracks requests to start nav transactions when a SentryNavEffect is created even if it doesn't currently have the lease. If the lease is later acquired and the transaction slot is available on the shared IScope, the staged transaction is started, which keeps ViewPager2 happy.
@0xadam-brown
0xadam-brown enabled auto-merge (squash) November 4, 2026 12:16
@0xadam-brown
0xadam-brown merged commit a21cd09 into main Nov 4, 2026
44 of 45 checks passed
@0xadam-brown
0xadam-brown deleted the 0xadam-brown/collection/nav3-deferred-transactions branch November 4, 2026 12:25
sentry-junior Bot added a commit that referenced this pull request Nov 5, 2026
Scope.setTransaction now passes transaction.getSpanContext().withDefaultStatus()
to scope observers (#6418 on 9.x.x). The nav3 tests from main (#6398) bind bare
ITransaction mocks whose getSpanContext() returns null, so they hit an NPE.

Co-Authored-By: Roman Zavarnitsyn <roman.zavarnitsyn@sentry.io>
sentry-junior Bot added a commit that referenced this pull request Nov 5, 2026
Scope.setTransaction now passes transaction.getSpanContext().withDefaultStatus()
to scope observers (#6418 on 9.x.x). The nav3 tests from main (#6398) bind bare
ITransaction mocks whose getSpanContext() returns null, so they hit an NPE.

Co-Authored-By: Roman Zavarnitsyn <roman.zavarnitsyn@sentry.io>
@sentry-junior sentry-junior Bot mentioned this pull request Nov 5, 2026
0xadam-brown added a commit that referenced this pull request Nov 13, 2026
…gment apps (JAVA-785, JAVA-823) (#6480)

This commit constitutes part 3 of our Nav3 support for multi-Activity and multi-Fragment apps.

In part 1 (#6301) we provided the basic mechanism for coordinating multiple SentryNavEffects; and in part 2 (#6398) we adapted our approach to the quirks of ViewPager2 and other oddball Fragment lifecycles.

Both of those commits assumed that any Activities or Fragments that had a SentryNavEffect would only launch other Activities or Fragments that also had SentryNavEffects. This commit handles three remaining (and important) corner cases:

1. Ensuring spans don't get lost when the user returns to an Activity or Fragment they already visited. (We now start a new nav transaction when the user returns if the previous transaction finished in the meantime.)

2. Ensuring Activities or Fragments with SentryNavEffects play nicely with Activities or Fragments without them. (We no longer keep stale nav data in the shared IScope.)

3. Ensuring Activities' ui.load transactions interoperate well with Nav3 transactions. (Nav3 transactions now consistently defer to ui.load transactions, like we do for Nav2.)

See the PR description of #6480 for more details.
runningcode added a commit that referenced this pull request Nov 17, 2026
* feat(android): Connect traces with web content in a WebView (#6395)

* feat(android): Connect traces with web content in a WebView

Add SentryWebViewTracing.addTracePropagation, which writes the current
trace into pages from allowed hosts as sentry-trace and baggage meta
tags before any page script runs, so @sentry/browser in the page
continues the trace on pageload.

It uses androidx.webkit's document-start script, a compileOnly
dependency checked at runtime, and adds a sample screen and an
instrumented test.

Refs JAVA-716

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Keep non-default ports and handle old androidx.webkit

A WebView origin rule without a port only matches port 443, so a
non-default port in an allowedHosts entry is now kept in the rule.

androidx.webkit is compileOnly, so an app may ship an older version
than the SDK was compiled against. A LinkageError is now caught, like
for androidx.activity in the feedback form, and the call returns false.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Remove the previous WebView script after an off-thread call

A call off the WebView's thread returns false without touching the
WebView, which left Sentry's previous script installed. The removal is
now posted to the WebView's thread, and skipped if a call on that
thread replaced the script in the meantime.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Track WebView script replacements per WebView

The off-thread cleanup compared against one counter shared by all
WebViews, bumped before the replacement finished. A concurrent
off-thread call could then remove a script that was just installed,
and a call on another WebView cancelled this WebView's cleanup. The
count is now kept per WebView, with weak keys, and bumped once the
replacement is done.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(samples): Move the WebView trace screen to the Tracing tab

The sample no longer shows an iframe; the instrumented test in
sentry-uitest-android still checks that iframes don't get the trace.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Match internationalized hosts in WebView trace propagation

The page's location.hostname is in punycode, so an allowedHosts entry
like münchen.de never matched. Non-ASCII entries are now converted
with IDN.toASCII, and entries that cannot be converted are dropped.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(changelog): Shorten WebView trace propagation entry

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Never throw from the posted WebView script removal

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Rethrow fatal errors from the posted WebView script removal

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Add docs link to the changelog

* Updated link to WebView documentation in CHANGELOG.

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(deps): Ignore OpenTelemetry Dependabot updates (#6531)

Prevent independent OpenTelemetry dependency bumps because the SDK, agent, alpha artifacts, and semantic conventions must be updated together.

Co-authored-by: Codex <noreply@openai.com>

* feat(android): Add WebView tracing options with allowed hosts (#6516)

* feat(android): Connect traces with web content in a WebView

Add SentryWebViewTracing.addTracePropagation, which writes the current
trace into pages from allowed hosts as sentry-trace and baggage meta
tags before any page script runs, so @sentry/browser in the page
continues the trace on pageload.

It uses androidx.webkit's document-start script, a compileOnly
dependency checked at runtime, and adds a sample screen and an
instrumented test.

Refs JAVA-716

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Keep non-default ports and handle old androidx.webkit

A WebView origin rule without a port only matches port 443, so a
non-default port in an allowedHosts entry is now kept in the rule.

androidx.webkit is compileOnly, so an app may ship an older version
than the SDK was compiled against. A LinkageError is now caught, like
for androidx.activity in the feedback form, and the call returns false.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Remove the previous WebView script after an off-thread call

A call off the WebView's thread returns false without touching the
WebView, which left Sentry's previous script installed. The removal is
now posted to the WebView's thread, and skipped if a call on that
thread replaced the script in the meantime.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Track WebView script replacements per WebView

The off-thread cleanup compared against one counter shared by all
WebViews, bumped before the replacement finished. A concurrent
off-thread call could then remove a script that was just installed,
and a call on another WebView cancelled this WebView's cleanup. The
count is now kept per WebView, with weak keys, and bumped once the
replacement is done.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(samples): Move the WebView trace screen to the Tracing tab

The sample no longer shows an iframe; the instrumented test in
sentry-uitest-android still checks that iframes don't get the trace.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Match internationalized hosts in WebView trace propagation

The page's location.hostname is in punycode, so an allowedHosts entry
like münchen.de never matched. Non-ASCII entries are now converted
with IDN.toASCII, and entries that cannot be converted are dropped.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(changelog): Shorten WebView trace propagation entry

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Never throw from the posted WebView script removal

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Rethrow fatal errors from the posted WebView script removal

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Add docs link to the changelog

* Updated link to WebView documentation in CHANGELOG.

* feat(android): Use tracePropagationTargets as the default WebView hosts

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* feat(android): Add WebView tracing options with allowed hosts

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Default WebView allowed hosts to an empty list

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(deps): bump retrofit from 2.9.0 to 3.0.0 (#6534)

Bumps `retrofit` from 2.9.0 to 3.0.0.

Updates `com.squareup.retrofit2:retrofit` from 2.9.0 to 3.0.0
- [Release notes](https://github.com/square/retrofit/releases)
- [Changelog](https://github.com/lysine-dev/retrofit/blob/trunk/CHANGELOG.md)
- [Commits](https://github.com/square/retrofit/compare/2.9.0...3.0.0)

Updates `com.squareup.retrofit2:converter-gson` from 2.9.0 to 3.0.0
- [Release notes](https://github.com/square/retrofit/releases)
- [Changelog](https://github.com/lysine-dev/retrofit/blob/trunk/CHANGELOG.md)
- [Commits](https://github.com/square/retrofit/compare/2.9.0...3.0.0)

---
updated-dependencies:
- dependency-name: com.squareup.retrofit2:retrofit
  dependency-version: 3.0.0
  dependency-type: direct:production
  update-type: version-update:semver-major
- dependency-name: com.squareup.retrofit2:converter-gson
  dependency-version: 3.0.0
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* docs(android): Fix Javadoc that Error Prone flags in sentry-android-core (#6498)

Error Prone never ran on sentry-android-core. Enabling it reports 34 Javadoc warnings, which
fail the build under -Werror:

- MissingSummary: turn the @return text into a "Returns ..." summary, and add summaries to
  Javadocs that only had block tags.
- InvalidBlockTag: replace the @noinspection unused Javadoc tags with @SuppressWarnings("unused"),
  which IntelliJ honors too.
- NotJavadoc: turn the Javadoc on methods of an anonymous class into regular comments.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android)!: Mark sentry-android-core classes final (#6494)

* ref(android)!: Mark sentry-android-core classes final

Error Prone's Nopen check requires every class to be final, abstract or @Open. It never ran on
sentry-android-core, so 42 classes were left open. None of them is subclassed except
ActivityLifecycleCallbacksAdapter, which becomes abstract since it is only extended, never
instantiated.

The protected members of the classes that are now final become package-private, as nothing can
subclass them anymore.

BREAKING CHANGE: AnrV2Integration, AnrProfilingIntegration, SentryUserFeedbackForm and its
Builder, WindowContentChangedCallback and several internal classes can no longer be subclassed.
ActivityLifecycleCallbacksAdapter can no longer be instantiated directly.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android)!: Adapt the final classes to main's deprecated feedback APIs

On main, the deprecated SentryUserFeedbackDialog and its Builder still
extend SentryUserFeedbackForm and its Builder, so those two stay @Open
until the dialog is removed. The deprecated SentryUserFeedbackButton and
SentryUserFeedbackDialog.Builder become final.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Don't block callers on network status registration (JAVA-828) (#6537)

AndroidConnectionStatusProvider no longer makes callers wait while it registers its network and telephony callbacks with system services. That's good because one of those callers is Sentry.init.

THE PROBLEM

Before this commit, AndroidConnectionStatusProvider used one lock to guard three code paths: callback registration, connection cache refreshes, and adding or removing observers. The SDK executor held that lock across ConnectivityManager.registerDefaultNetworkCallback and, on API 31+, TelephonyManager.registerTelephonyCallback. Both are calls into system services that can occasionally block for seconds.

The main thread waited on the same lock in two places:

Session Replay adding its observer during Sentry.init.
Every captured event reading the connection type or status.
If the system service registrations took a while, the main thread would stall. Scenario (1) in particular could lead to ANRs during app startup. Oof.

A ROAD NOT TAKEN 

Any fix has to isolate the slow part (registration) so that the main thread can return quickly when reading cached network info or adding observers.

The narrowest approach would be to give registration its own lock and keep a second lock for the cache and observers. But a refresh would then either hold the cache lock while querying Android, making readers wait on that query, or need a version counter that every write must remember to update. Even worse, two locks would need to document order they're taken in to avoid deadlock, which would be easy to get wrong.

A BETTER APPROACH

The approach we ultimately landed on was to avoid having the main thread hold a lock entirely, and instead rely on atomically publishing cached network info (see commit 2). We did so by introducing a new ConnectionSnapshot class, which is immutable and can be safely shared across threads. We stored observers in copy-on-write lists rather than having them be explicitly locked, and we confined registration to the the SDK executor (it's still defensively guarded by its own lock).

Callers that need to access cached network info or register observers can now return quickly. Either of approach would have allowed as much, of course, but the benefit of introducing the snapshot it that it also:

- lets us consolidate a lot of related state that the provider previously had to handle separately
- avoids the pain of managing (yet another) lock
- lays the groundwork for future PRs that confine refreshes to the SDK executor and let snapshots tell a lost network from capabilities that are simply unknown.

Co-authored-by: Lazizbek Ergashev <20501725+lazerg@users.noreply.github.com>

* build: Upgrade Gradle to 9.9.0 (#6539)

The wrapper jar was regenerated by the 9.9.0 wrapper task and matches
the official 9.9.0 wrapper checksum.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(deps): bump room2 from 2.8.4 to 2.8.5 (#6550)

Bumps `room2` from 2.8.4 to 2.8.5.

Updates `androidx.room:room-compiler` from 2.8.4 to 2.8.5

Updates `androidx.room:room-ktx` from 2.8.4 to 2.8.5

Updates `androidx.room:room-runtime` from 2.8.4 to 2.8.5

---
updated-dependencies:
- dependency-name: androidx.room:room-compiler
  dependency-version: 2.8.5
  dependency-type: direct:production
  update-type: version-update:semver-patch
- dependency-name: androidx.room:room-ktx
  dependency-version: 2.8.5
  dependency-type: direct:production
  update-type: version-update:semver-patch
- dependency-name: androidx.room:room-runtime
  dependency-version: 2.8.5
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build(android): Bump AGP to 9.5.1 (#6540)

Raise the default Android Gradle Plugin from 9.4.0 to 9.5.1, the latest
stable release. The AGP matrix moves to 9.0.0/9.4.0/9.5.1, keeping the
minimum supported version and the previous default, and the replay
snapshot upload follows the new default.

Lint 9.5 can't load the custom checks bundled with older
androidx.lifecycle releases and reports ObsoleteLintCustomCheck, which
warningsAsErrors fails on. Disable it in the affected modules; the
skipped checks cover lifecycle APIs those modules don't call.

Lint 9.5's new ThreadConstraint check flags
ScreenshotUtils.takeScreenshot, which is safe to call from any thread
because captureScreenshot posts the view draw to the UI thread. Suppress
it there.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* fix(replay): Measure gesture batch durations on a monotonic clock (JAVA-830) (#6538)

* fix(replay): Measure gesture batch durations on a monotonic clock

Touch moves are batched and sent as one RRWebInteractionMoveEvent every 500ms. Both the batch
timer and each position's timeOffset subtracted wall-clock readings, so a clock step corrupted
them: a backward step held the batch back until the clock caught up, and dropped it if the finger
lifted first; a forward step sent it at once with offsets stretched by the step.

Both now come from one Stopwatch on the converter's MonotonicTicker, started with the first
batched position. Without a clock step the output is unchanged. The event timestamps stay on the
wall clock; they move to an AnchoredClock in v9.

Fixes JAVA-830
Fixes #6523

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* fix(compose): Detect lazy layout scrolls on Compose foundation 1.10+ (#6548)

Compose foundation 1.10 replaced ScrollingContainerElement with
ScrollableAreaElement, so the gesture target locator no longer found
LazyColumn/LazyRow/LazyGrid/Pager as scroll targets and dropped the
scroll breadcrumb. Match ScrollableAreaElement and keep its class name
under R8.

Fixes #6547

---------

Co-authored-by: sentry-junior[bot] <264270552+sentry-junior[bot]@users.noreply.github.com>
Co-authored-by: Gene Wright <gene.wright@sentry.io>

* build: Upgrade Gradle to 9.9.1 (#6552)

The 9.9.1 wrapper task leaves gradle-wrapper.jar unchanged, and the jar
matches the official 9.9.1 wrapper checksum.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* build: Move OpenTelemetry agent test wiring into a convention plugin (#6551)

The Spring Boot 2, 3 and 4 modules each copied the setup that runs the agent tests in a child
JVM. The io.sentry.otel-agent-tests plugin now owns it, so the modules keep only the dependencies
of their Spring Boot line.

Fixes #6378
Fixes JAVA-801

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(deps): bump ktorClient from 3.0.0 to 3.6.1 (#6478)

* chore(deps): bump ktorClient from 3.0.0 to 3.6.1

Bumps `ktorClient` from 3.0.0 to 3.6.1.

Updates `io.ktor:ktor-client-core` from 3.0.0 to 3.6.1
- [Release notes](https://github.com/ktorio/ktor/releases)
- [Changelog](https://github.com/ktorio/ktor/blob/3.6.1/CHANGELOG.md)
- [Commits](https://github.com/ktorio/ktor/compare/3.0.0...3.6.1)

Updates `io.ktor:ktor-client-java` from 3.0.0 to 3.6.1
- [Release notes](https://github.com/ktorio/ktor/releases)
- [Changelog](https://github.com/ktorio/ktor/blob/3.6.1/CHANGELOG.md)
- [Commits](https://github.com/ktorio/ktor/compare/3.0.0...3.6.1)

---
updated-dependencies:
- dependency-name: io.ktor:ktor-client-core
  dependency-version: 3.6.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
- dependency-name: io.ktor:ktor-client-java
  dependency-version: 3.6.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>

* fix(ktor): [Ktor 3.6 2] Preserve Ktor 3.0 runtime compatibility (#6517)

* fix(ktor): Preserve Ktor 3.0 runtime compatibility

Avoid inlining Ktor 3.6's AttributeKey factory. The inlined code
references a TypeInfo constructor unavailable in older Ktor 3.x
runtimes.

Construct TypeInfo explicitly through its stable two-argument
constructor instead.

* test(ktor): [Ktor 3.6 3] Add Ktor 3.0 compatibility suite (#6520)

* test(ktor): Add Ktor 3.0 compatibility suite

Compile the production integration against the catalog Ktor version
while compiling and running the existing tests against Ktor 3.0.0.

Add the compatibility suite to check so inline API binary
incompatibilities fail CI.

* test(ktor): Simplify Ktor 3.0 suite configuration

Use Ktor's enforced platform to pin all compatibility test artifacts to
3.0.0 instead of maintaining a custom resolution strategy. Reuse the
standard test source directories so the suite follows source changes.

Co-Authored-By: Claude <noreply@anthropic.com>

* Update sentry-ktor-client/build.gradle.kts

Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Alexander Dinauer <adinauer@users.noreply.github.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>

* ref(android): Fix Error Prone warnings in sentry-android-core (#6499)

* ref(android): Fix Error Prone warnings in sentry-android-core

Error Prone never ran on sentry-android-core. Enabling it reports 59 warnings outside Javadoc,
which fail the build under -Werror.

- Suppress FutureReturnValueIgnored, JavaUtilDate, and ReferenceEquality where identity is
  intended, as the sentry module does.
- Keep the AnrProfilingIntegration updater a field, since removeCallbacks needs the instance that
  was posted, and suppress UnnecessaryLambda there.
- Read the ANR thread dump and the app start profiling config as UTF-8, Android's default charset.
- Remove unused fields and parameters, add missing @Override, make long-to-double conversions
  explicit, and narrow members that are effectively private or protected in a final class.
- Explain the two catch blocks that ignore their exception.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Fix the Error Prone warnings only main has

- ANRWatchDog.monotonicTicker is never read (#6041), since the ticker lambda captures the
  constructor parameter. Remove the field and keep the parameter.
- ActivityFramesTracker.lock is protected in a final class; make it package-private.
- LifecycleWatcher reads the session start Date to seed the session window; suppress JavaUtilDate
  there, as the other Date uses are.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Keep the default charset reads as they were

Android's default charset is always UTF-8, so DefaultCharset is
disabled for the Android modules in #6493 instead of being fixed here.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Drop the UnnecessaryLambda suppressions

#6493 disables UnnecessaryLambda for the Android modules, so the
suppressions on the ANRWatchDog and AnrProfilingIntegration fields
are redundant.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Drop the updater field comment

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(android): Make the local-class method comment a plain comment

Error Prone's NotJavadoc flags Javadoc on methods of local classes, which
#6537 added in AndroidConnectionStatusProvider.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Resolve NullAway findings in sentry-android-core (#6497)

* fix(android): Resolve NullAway findings in sentry-android-core

NullAway never ran on sentry-android-core. Enabling it reports 32 errors. Most are fields and
return types that are nullable but not annotated, and fields NullAway cannot see as checked
because a check happens in a different scope (a lambda, another method) than the dereference.

- Annotate the nullable fields and return types, including ViewHierarchyNode.setType and
  UiElement.equals in the sentry module.
- Read options once in lambdas, and pass scopes and options to
  createStandaloneAppStartTransaction, whose callers already check them.
- Compute the device family once instead of through a LazyEvaluator, whose evaluator must not
  return null.
- Create the system events HandlerThread in the branch that uses it, inside the existing try.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Build the system events IntentFilter where it is registered

The filter was a nullable field, built on the first registration and cleared in close(). It
only depends on the final actions array, so build it in the registration task instead. That
leaves nothing nullable to track.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Check options before finishing the app start span

finishAppStartSpan passed the nullable options field to getAppStartTimeSpanWithFallback, which
NullAway reports on main. The end time is only used when performance is enabled, which register()
sets together with options, so return early when either is unset.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Annotate the view hierarchy byte array as nullable

@Nullable byte[] annotates the element type for TYPE_USE annotations, so
Kotlin saw ByteArray instead of ByteArray?.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* build: Stop logging passed tests and test output to the console (#6554)

Every passing test printed a PASSED line plus its stdout/stderr, which
buried the failures in CI logs. The console now shows only skipped and
failed tests. Test output is still in the HTML and XML test reports.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* feat(android): Add experimental out-of-process crash reporting (#6347)

* feat(android): Add service-managed NDK crash daemon

Run the NDK crash daemon from an isolated Android service process so it can
release the crashing app, merge the resulting tombstone, and send the
enriched envelope before shutdown.

* Update CHANGELOG.md

* fix(android): Narrow exception handling in the NDK crash service

Replace the broad `Throwable` catches in `SentryNdkCrashService` with
the exceptions the guarded code can throw, and restore the interrupt
flag when the service bind is interrupted.

* fix(android): Stop the NDK crash service when its daemon exits

The service was started with `startService` but never stopped, so its
process kept running after `Sentry.close()` or after the app process
died. Call `stopSelf` with the start ID that was current when the
worker started, so a newer start from a relaunched app is not stopped.

* fix(android): Match the NDK crash service stop to its own start

A new service process receives `onBind` before `onStartCommand`, so a
worker started from the binder configuration could capture a stale
start ID and never stop the service. Tag each start with its shared
memory path and stop only when the latest start belongs to the
exiting worker.

* fix(android): Write the NDK tombstone marker after a successful send

The marker was written as soon as the merged envelope existed. If the
send then failed with a network error, the daemon discarded the
envelope and the next launch skipped the tombstone as already
reported, so the crash was lost. Write the marker only once the send
succeeds.

* fix(android): Bind the NDK crash service for the app's lifetime

`startService` throws when the app starts in the background, and the
started service stopped about a minute after the app left the
foreground, so a later crash found no daemon. Native init then fell
back to fork/exec, which cannot exec `libsentry-crash.so` from inside
the APK with default packaging, and stalled startup for 10 s.

Bind the service with `BIND_AUTO_CREATE` and keep the binding until
`SentryNdk.close()`, so the service process keeps the app's importance
and is not frozen. Let crash processing continue after `onDestroy`,
which runs as soon as a crashed app's binding goes away. Fail NDK init
instead of falling back to fork/exec when the service is unavailable:
below API 30, without a context, or when binding or configuring fails.

Move the inproc `SentryNdkTest` to inproc-only test sources, since
outproc init now requires a context.

* fix(android): Match NDK tombstones by the app's pid

The crash service picked the native crash exit info closest to the
event's timestamp and ignored its pid. In a crash loop, the first poll
can run before the current crash has an exit info, and the previous
crash's tombstone was merged instead.

Skip exit infos of other pids. The worker captures the pid when it
starts, so a later configure call cannot change it.

* fix(android): Start an NDK crash daemon for each relaunch

When the app relaunched while the previous daemon was still sending
its crash, or before it noticed that its process was gone, the new
config was stored but no worker started, and the old worker's cleanup
then cleared it. The relaunched app's next crash waited 10 s in the
handler and was lost, and its descriptors leaked.

Keep the per-app daemon state in a config owned by its worker, and
start a worker for each config right away. The old daemon is past
crash processing or only waiting for its process to go away, so the
two share only process-wide native state such as the logger.

* ci(temp): Publish sentry-native-ndk-outproc to Maven Local

`sentry-android-ndk-outproc` depends on
`io.sentry:sentry-native-ndk-outproc`, which is not on Maven Central
yet. Build it from the sentry-native feature branch and publish it to
Maven Local so the Build job can resolve it.

TODO: Revert. This is temporary until sentry-native publishes the artifact.

* fix(android): Wait for the NDK tombstone outside the send budget

The tombstone wait and the send shared `shutdownTimeoutMillis`, so a
slow tombstone left the send no time at all.

Wait for the app's exit info instead. The app exits right after
debuggerd writes its tombstone, so the wait ends when the tombstone is
there, capped at the 30 s match tolerance. Stop early when the app did
not crash natively or the tombstone does not parse. Give the send the
full `shutdownTimeoutMillis`, and write the marker for a merged
envelope that was sent or kept for the next launch.

* fix(android): Remove stale NDK crash IPC files

Name the shared-memory file after the app pid and delete leftovers whose
process no longer runs when the crash service starts. A setup that fails
before the daemon attaches would otherwise leave the file behind, while
other processes of a multi-process app keep theirs.

* fix(android): Fall back to the 1-argument NDK init

With a context, NdkIntegration only looked up
`SentryNdk.init(Context, SentryAndroidOptions)`. An older
sentry-android-ndk without it disabled NDK. Fall back to
`SentryNdk.init(SentryAndroidOptions)` and log a warning.

* ref(android): Stop passing the tombstone flag to NdkOptions

sentry-native-ndk-outproc no longer reads it, because it always
sends native crash envelopes without a minidump. The crash service
still gets the tombstone flag through its configure call.

* fix(android): Skip SDK init in the NDK crash service process

The crash service runs in its own `.sentry_ndk_crash` process, which
runs the app's `Application.onCreate`. Apps that call
`SentryAndroid.init` there started a second SDK on every launch, with
the tombstone, ANR and outbox senders working on the same cache
directory as the app process.

Return early from `SentryAndroid.init` in that process.

* ref(android): Remove the NDK crash service process check

SentryAndroid.init now returns early in the `.sentry_ndk_crash`
process, so SentryNdk.init is no longer called there.

* fix(android): Keep a newer NDK tombstone marker

Crash service workers for processes that crash close together can
finish out of order. A worker that wrote an older tombstone timestamp
let `TombstoneIntegration` report the newer crash again on the next
launch.

Write the marker only when it is newer than the current one, and
serialize the workers.

* ci: Stop the sentry-native Gradle daemon after publishing

The publish step left its 4 GB Gradle daemon running alongside the
8 GB daemon used by `make preMerge`. The job was killed about 12
minutes into the tests, most likely because the runner ran out of
memory.

* fix(android): Bind the NDK crash service as important

Without `BIND_IMPORTANT`, a foreground app raises its bound crash
service process only to the visible oom adj (100), so lmkd would kill
the service before the app under memory pressure. The flag gives the
service the app's own adj (0) while the app is on top, without
lowering the app's priority the way `BIND_ABOVE_CLIENT` would.

* Apply batched suggestions from code review

Co-authored-by: Roman Zavarnitsyn <rom4ek93@gmail.com>

* chore(android): Update the outproc API dump

The outproc module shares NdkScopeObserver with the inproc module,
which gained setEnvironment on main (#5772).

* fix(android): Reattach the NDK crash daemon after a service restart

When the crash service process died while the app was bound, Android
restarted it and called `onServiceConnected` again, but the connection
had already closed the descriptors it configured with, so the app ran
without a crash daemon. Configure the restarted service with new pipes
and the app's shared memory file, then reattach the native crash IPC.
Bind again when the binding dies, and log a lost connection at ERROR.

* ref(android): Build the NDK outproc package on top of sentry-android-ndk

sentry-android-ndk-outproc compiled sentry-android-ndk's sources
through extra source folders, shared its namespace and borrowed its
ProGuard rules, so the same files belonged to both modules, and the
deprecated `BaseExtension` wired it up. Depend on sentry-android-ndk
instead, without its `sentry-native-ndk`, and keep only the crash
service and the `SentryNdkOutproc` entry point in the
`io.sentry.android.ndk.outproc` namespace. `NdkIntegration` now loads
that entry point first and falls back to `SentryNdk`.

* fix(android): Write the NDK tombstone marker before the send

The native side keeps the envelope for the next launch until it is
sent, so the crash is delivered whatever the send result. Write the
marker before the send, so that a send that a frozen service finishes
late no longer races TombstoneIntegration on the next launch.

* fix(android): Close rejected NDK crash service fds

`configureDaemon` left the file descriptors of a binder transaction
open until garbage collection when it rejected the configuration.
Close them before returning.

* fix(android): Time the NDK tombstone poll with a monotonic clock

The crash service polled for a matching tombstone until a wall clock
deadline, so a clock step backward delayed the crash send by the size
of the step. Use `SystemClock.uptimeMillis()`, which also drives
`Thread.sleep()`.

* fix(android): Narrow the NDK outproc library load catch

The library loader thread swallowed every `Throwable`, including fatal
errors. Catch only the `UnsatisfiedLinkError` of a missing or broken
library, which `init()` reports later. `ExceptionUtils.rethrowIfFatal`
does not fit, because it rethrows `LinkageError` and would turn a
missing library into an app crash.

* fix(android): Load the NDK outproc libraries before binding the service

A missing or broken native library surfaced only in `initCrashDaemon`,
after the crash service was bound and the IPC fds were detached. The
`UnsatisfiedLinkError` left the service bound with `BIND_IMPORTANT`
and the fds open for the app's lifetime. Load the libraries first, so
the failure leaves nothing behind.

* fix(android): Write the NDK tombstone marker when the merge fails

When the crash service matched the app's native crash exit but could
not merge its tombstone, because it stayed unavailable until the
deadline or the merge threw, it sent the native event without writing
the marker. On the next launch, `TombstoneIntegration` then reported
the same crash again as a standalone tombstone event. Write the marker
for a matched tombstone, merged or not.

* docs(android): Remove the Gradle plugin from the NDK outproc README

The `ndkBackend` option of the Gradle plugin is parked for now, so the
outproc package is selected through its dependency only.

* fix(android): Hand the NDK crash memfd to the crash service

The NDK now backs the crash IPC with a memfd instead of a file, because
the sparse file could raise `SIGBUS` on a full disk.

Create the memfd before configuring the service and send it over
Binder, which reattaching already does. This removes the file path from
the protocol, and the cleanup of stale files.

* docs(changelog): Move the outproc entry back to Unreleased

The merge of `origin/main` placed it under `8.63.0`, which was released
without out-of-process crash reporting.

* build(deps): Bump sentry-native-ndk-outproc to 1.0.0-beta.1

sentry-native now publishes `sentry-native-ndk-outproc` to Maven
Central, so CI no longer builds it from the sentry-native branch into
Maven Local.

* meta(android): Register the NDK outproc module for release

Craft's `registry` target fails the release when a published module
has no entry. Also list the module in the README packages table and
the Android issue template.

* ref(android): Pass the thread id as the NDK crash daemon's appTid

`appTid` received `System.nanoTime()`, but the daemon expects the
parent thread id. On Android it only names the shared memory file that
the daemon unlinks, and none exists because the memory is a memfd.

* ref(android): Time the NDK tombstone poll with a Deadline

New code reads clocks through `io.sentry.time` instead of `SystemClock`.
The poll deadline now comes from the options' monotonic ticker, which
also counts time the device spends suspended.

* ref(android): Document the NDK crash service's broad catches

The exception rules in `AGENTS.md` allow a broad catch only at a
boundary, with a comment on why and what happens on failure. The two
`Exception` catches also call `ExceptionUtils.rethrowIfFatal`, which
restores the interrupt flag of a swallowed `InterruptedException`.

* fix(android): Unbind the NDK crash service on any native init failure

`SentryNdkOutproc.init` only unbound the crash service when native init
threw `IllegalStateException`. Any other throwable, such as an
`UnsatisfiedLinkError`, left the service bound with `BIND_IMPORTANT`,
and the disabled NDK integration never closed it.

* docs(changelog): Explain how to enable Android outproc crash reporting

---------

Co-authored-by: Roman Zavarnitsyn <rom4ek93@gmail.com>

* test(android): Use fewer Robolectric SDK levels in core and replay (#6544)

* collection: Robolectric CI speed-up

* test(android): Use fewer Robolectric SDK levels in core and replay

Robolectric starts a new sandbox for each SDK level a test JVM uses, and downloads one android-all jar per level. Move pinned levels to the nearest shared level that takes the same code branch: 23/25 -> 24, 26 -> 30, 31/34 -> 33. Level 28 stays because ReplaySmokeTest and AndroidConnectionStatusProviderTest behave differently at 30.

Core and replay now use 6 levels instead of 11. Locally, core Robolectric tests go from about 82s to 71s.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(docs): Remove OkHttp version bump from changelog and add known issue for #6537 (#6561)

* build(android): Run Error Prone and NullAway on the Android modules (#6493)

* build(android): Run Error Prone and NullAway on the Android modules

net.ltgt.errorprone dropped Android support in 3.0.0, which we have used since 6.7. Since then
it only enables Error Prone and adds it to the processor path for Gradle's JVM source sets. On
sentry-android-core and sentry-uitest-android the errorprone dependencies and the NullAway ERROR
setting had no effect, and the Java compile tasks ran plain javac.

Use the AGP variant API to add the errorprone configuration to each variant's annotation
processor configuration and enable Error Prone on its JavaCompile task. Both APIs exist in AGP
9.0.0, the oldest version in the AGP matrix.

Pass -XDandroidCompatible=true so Error Prone treats these modules as Android code: it skips
checks that don't apply there, such as DefaultCharset, and runs its Android checks.

sentry-uitest-android has no hand-written Java, so exclude the view binding classes AGP
generates, which trip LabelledBreakTarget.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Resolve NullAway findings in AndroidConnectionStatusProvider

NullAway treats AtomicReference.get() as nullable, so the snapshot
reads added in #6537 fail once NullAway runs on sentry-android-core.
The snapshot is never set to null, so read it through one accessor
that suppresses the finding.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* release: 8.64.0

* build(android): Download Robolectric android-all jars with Gradle (#6553)

Robolectric fetched ~0.95 GB of android-all jars from Maven Central while
tests ran, and CI doesn't cache ~/.m2, so every Build run downloaded them
again. A new io.sentry.robolectric build-logic plugin resolves them with
Gradle before the tests start, so setup-gradle restores them from cache,
and points Robolectric at them through robolectric-deps.properties
instead of copying them into robolectric.dependency.dir.

See https://robolectric.org/blog/2023/11/11/improving-android-all-downloading/

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* feat(profiling): Store the thread state in ANR stack samples (#6587)

* ref(android): Store the thread state in ANR stack samples

* ref(android): Mark a null ANR sample state explicitly and document the format

* ref(android): Remove narrating comments from AnrStackTrace

* chore(docs): Add version for fix under Known Issues entries in changelog (#6569)

* fix(android): Measure debounce windows on the monotonic ticker (JAVA-729) (#6590)

* fix(android): Measure debounce windows on the monotonic ticker (JAVA-729)

Debouncer read AndroidCurrentDateProvider, which is
SystemClock.uptimeMillis() and stops in deep sleep. A window that
expired while the device slept still looked open, so the first
low-memory or battery breadcrumb after a long Doze could be dropped.

Debouncer now measures its window with a Deadline on
options.getMonotonicTicker(), which is CLOCK_BOOTTIME on Android. The
screenshot, view hierarchy, system events and app components debouncers
all use it.

AndroidEnvelopeCache still uses AndroidCurrentDateProvider: it compares
against TimeSpan's uptime start, so it can't move on its own.

Refs JAVA-729

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* ref(android): Remove unused BreadcrumbFactory (#6598)

* perf(core): Avoid allocations in log calls when debug is off (#6600)

* perf(core): Avoid allocations in log calls when debug is off

Every SDK log call allocated a varargs array at the call site, also when debug is off. ILogger now has fixed-arity default overloads for zero, one and two arguments, which DiagnosticLogger and NoOpLogger override. Recompiled call sites use them and do not allocate when logging is disabled.

Hot-path log calls in DefaultCompositePerformanceCollector and SentryGestureListener also built their messages eagerly with string concatenation or toString(). They now pass format arguments, so the message is built only when logging is enabled.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog: Narrow the allocation claim

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* fix(android): Don't swallow interrupts while flushing a startup crash (#6568)

* fix(android): Don't swallow interrupts while flushing a startup crash

After a startup crash, SendCachedEnvelopeIntegration blocks the thread that calls Sentry.init
until the cached envelopes are sent. If that thread was interrupted, future.get() cleared the
interrupt and the trailing catch (Throwable) swallowed it, so the caller never saw it. It also
logged that cached events will not be sent, although the task keeps sending them.

Catch InterruptedException next to the timeout, restore the interrupt and continue in the
background.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* feat(replay): Add pauseTouchTracking and resumeTouchTracking (#6546)

Apps can stop recording touches on sensitive screens, such as a custom PIN pad, while
the masked screen recording continues. Masking hides the content, but recorded touch
positions could reveal the input.

The flag is independent of pause()/resume() and is cleared only by
resumeTouchTracking(), also across new replays. It is set directly instead of being queued
on the main thread, so it applies to touch events that are already pending.

Also mark debugMaskingEnabled as volatile, it is written and read on different threads.

Fixes JAVA-658

Fixes #5787

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* test(okhttp): Wait until the reader is blocked before closing the body (#6604)

The concurrent close test signaled right after the reader got the first chunk, before it
entered the next read(). When close() landed in that gap, the reader's next read() threw
Okio's "closed" IllegalStateException, the thread died, and the test timed out waiting for
it. ParkingSource now signals right before it blocks, so close() always lands while the
reader is inside read().

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

* fix(android): Part 3 of Nav3 support for multi-Activity and multi-Fragment apps (JAVA-785, JAVA-823) (#6480)

This commit constitutes part 3 of our Nav3 support for multi-Activity and multi-Fragment apps.

In part 1 (https://github.com/getsentry/sentry-java/pull/6301) we provided the basic mechanism for coordinating multiple SentryNavEffects; and in part 2 (https://github.com/getsentry/sentry-java/pull/6398) we adapted our approach to the quirks of ViewPager2 and other oddball Fragment lifecycles.

Both of those commits assumed that any Activities or Fragments that had a SentryNavEffect would only launch other Activities or Fragments that also had SentryNavEffects. This commit handles three remaining (and important) corner cases:

1. Ensuring spans don't get lost when the user returns to an Activity or Fragment they already visited. (We now start a new nav transaction when the user returns if the previous transaction finished in the meantime.)

2. Ensuring Activities or Fragments with SentryNavEffects play nicely with Activities or Fragments without them. (We no longer keep stale nav data in the shared IScope.)

3. Ensuring Activities' ui.load transactions interoperate well with Nav3 transactions. (Nav3 transactions now consistently defer to ui.load transactions, like we do for Nav2.)

See the PR description of https://github.com/getsentry/sentry-java/pull/6480 for more details.

* chore(deps): Scan one Gradle directory with Dependabot (#6603)

The separate sample and test builds read the SDK build's version catalog, so listing them as Dependabot directories opened one PR per directory for the same catalog bump. The root directory covers the catalog; the three versions only the samples declared move into it.

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>

* build(deps): Bump OpenTelemetry to 2.33.1 (#6602)

* build(deps): Bump OpenTelemetry to 2.33.1

Align the OpenTelemetry instrumentation and Java agent with SDK 1.67.0. Update tests for the SDK optimization that returns non-recording spans when no span processor is configured.

Co-Authored-By: Claude <noreply@anthropic.com>

* changelog

* changelog

* build(deps): Align OpenTelemetry semantic conventions

Use the semantic convention versions tested with OpenTelemetry Java instrumentation 2.33.1.

Co-Authored-By: Claude <noreply@anthropic.com>

* docs(changelog): Clarify OpenTelemetry dependency bump

List the semantic convention upgrades and explain how missing span processors affect outgoing Sentry trace headers.

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>

* build(deps): Prepare for Spring Boot 4.2 (#6532)

* build(deps): Bump Spring Boot to 4.2.0-RC2

Align OpenTelemetry with the version managed by Spring Boot 4.2. Keep RestTemplate support compiling despite upstream removal warnings and update tests for OpenTelemetry's no-processor optimization.

Co-Authored-By: Claude <noreply@anthropic.com>

* changelog

* fix(spring): Suppress RestTemplate removal warnings

Keep the Spring 7 sample compiling with warnings treated as errors after the Spring Boot 4.2 dependency bump.

Co-Authored-By: Claude <noreply@anthropic.com>

* build(deps): Add Spring milestone repository

Resolve Spring Boot 4.2 release candidate plugin and framework artifacts in clean CI environments.

Co-Authored-By: Claude <noreply@anthropic.com>

* test(spring): Build valid GraphQL operation documents

Include a selection set so GraphQL Java 25 can print the test operations used by Spring subscription error reporting tests.

Co-Authored-By: Claude <noreply@anthropic.com>

* build(deps): Bump OpenTelemetry to 2.33.1

Align the OpenTelemetry instrumentation and Java agent with SDK 1.67.0. Update tests for the SDK optimization that returns non-recording spans when no span processor is configured.

Co-Authored-By: Claude <noreply@anthropic.com>

* changelog

* build(spring): Keep Spring Boot 4.1 dependency

Limit the preparation PR to forward-compatible source and test changes. Remove the Spring Boot 4.2 release candidate bump, milestone repositories, and dependency changelog entry.

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>

* chore(deps): bump org.hsqldb:hsqldb from 2.6.1 to 2.7.4 (#6572)

Bumps org.hsqldb:hsqldb from 2.6.1 to 2.7.4.

---
updated-dependencies:
- dependency-name: org.hsqldb:hsqldb
  dependency-version: 2.7.4
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* chore(deps): bump the github-actions group across 1 directory with 4 updates (#6610)

Bumps the github-actions group with 4 updates in the / directory: [getsentry/github-workflows/danger](https://github.com/getsentry/github-workflows), [getsentry/craft](https://github.com/getsentry/craft), [getsentry/github-workflows/updater](https://github.com/getsentry/github-workflows) and [getsentry/github-workflows/validate-pr](https://github.com/getsentry/github-workflows).


Updates `getsentry/github-workflows/danger` from 3.4.1 to 3.5.0
- [Release notes](https://github.com/getsentry/github-workflows/releases)
- [Changelog](https://github.com/getsentry/github-workflows/blob/main/CHANGELOG.md)
- [Commits](https://github.com/getsentry/github-workflows/compare/959162c159b866489d3fabe93327f48f9b8cb7ee...36529d71129a14967bea1fb81bcaf2a5943cf877)

Updates `getsentry/craft` from 3.11.2 to 3.12.0
- [Release notes](https://github.com/getsentry/craft/releases)
- [Changelog](https://github.com/getsentry/craft/blob/master/CHANGELOG.md)
- [Commits](https://github.com/getsentry/craft/compare/9567926a606072678ef367b02445c4f99f374795...0da9309228741a4c300d3c87ed7dcafa8f703525)

Updates `getsentry/github-workflows/updater` from 3.4.1 to 3.5.0
- [Release notes](https://github.com/getsentry/github-workflows/releases)
- [Changelog](https://github.com/getsentry/github-workflows/blob/main/CHANGELOG.md)
- [Commits](https://github.com/getsentry/github-workflows/compare/959162c159b866489d3fabe93327f48f9b8cb7ee...36529d71129a14967bea1fb81bcaf2a5943cf877)

Updates `getsentry/github-workflows/validate-pr` from 3.4.1 to 3.5.0
- [Release notes](https://github.com/getsentry/github-workflows/releases)
- [Changelog](https://github.com/getsentry/github-workflows/blob/main/CHANGELOG.md)
- [Commits](https://github.com/getsentry/github-workflows/compare/959162c159b866489d3fabe93327f48f9b8cb7ee...36529d71129a14967bea1fb81bcaf2a5943cf877)

---
updated-dependencies:
- dependency-name: getsentry/github-workflows/danger
  dependency-version: 3.5.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: github-actions
- dependency-name: getsentry/craft
  dependency-version: 3.12.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: github-actions
- dependency-name: getsentry/github-workflows/updater
  dependency-version: 3.5.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: github-actions
- dependency-name: getsentry/github-workflows/validate-pr
  dependency-version: 3.5.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: github-actions
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* chore(deps): bump com.github.ben-manes.versions from 0.42.0 to 0.67.0 (#6573)

Bumps com.github.ben-manes.versions from 0.42.0 to 0.67.0.

---
updated-dependencies:
- dependency-name: com.github.ben-manes.versions
  dependency-version: 0.67.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build(deps): Bump Spotless to 8.11.0 (#6612)

* build(deps): Bump Spotless to 8.11.0

Spotless 8.11.0 bumps the default google-java-format from 1.28.0 to 2.0.0 and
ktfmt from 0.63 to 0.65, so this also reformats the codebase with make format.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* build(android): Update detekt baseline for ktfmt trailing commas

Detekt baseline IDs embed the flagged code, so the trailing commas ktfmt 0.65 adds
no longer matched two entries in sentry-android-distribution.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* ref(distribution): Use check, require and error instead of throwing

Fixes the UseCheckOrError and UseRequire detekt issues instead of keeping them
in the baseline. The thrown exception types and messages stay the same.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* ci: Add PR risk (Jev) workflow (#6615)

* ci: Add PR risk (Jev) workflow

Copy getsentry/toolkit's pr-risk-jev.yml, which labels non-draft PRs
risk: low/medium/high via getsentry/pr-risk-action. Pin
actions/upload-artifact to the SHA the other workflows here use.

Co-Authored-By: Nelson Osacky <nelson.osacky@sentry.io>

* ci: Bump pr-risk-action and align github-script pin

Pin pr-risk-action to the latest main commit, which bumps Node to
22.23.4 for a security fix. Use the github-script v9.0.0 SHA the
other workflows in this repo already use.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: sentry-junior[bot] <264270552+sentry-junior[bot]@users.noreply.github.com>
Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* build(deps): Bump Room 3 to 3.0.3 (#6616)

Room 3 is now stable, so move off 3.0.0-rc01. Room 3.0.3 only needs
androidx.sqlite 2.7.x, so the sample's SQLite drivers go back from
2.8.0-alpha03 to stable 2.7.1. The sqliteRc catalog key is renamed to
sqliteDrivers since it no longer tracks an RC, which also stops
Dependabot from picking pre-releases for it.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(deps): bump coroutines from 1.6.1 to 1.11.0 (#6609)

Bumps `coroutines` from 1.6.1 to 1.11.0.

Updates `org.jetbrains.kotlinx:kotlinx-coroutines-core` from 1.6.1 to 1.11.0
- [Release notes](https://github.com/Kotlin/kotlinx.coroutines/releases)
- [Changelog](https://github.com/Kotlin/kotlinx.coroutines/blob/master/CHANGES.md)
- [Commits](https://github.com/Kotlin/kotlinx.coroutines/compare/1.6.1...1.11.0)

Updates `org.jetbrains.kotlinx:kotlinx-coroutines-android` from 1.6.1 to 1.11.0
- [Release notes](https://github.com/Kotlin/kotlinx.coroutines/releases)
- [Changelog](https://github.com/Kotlin/kotlinx.coroutines/blob/master/CHANGES.md)
- [Commits](https://github.com/Kotlin/kotlinx.coroutines/compare/1.6.1...1.11.0)

Updates `org.jetbrains.kotlinx:kotlinx-coroutines-test` from 1.6.1 to 1.11.0
- [Release notes](https://github.com/Kotlin/kotlinx.coroutines/releases)
- [Changelog](https://github.com/Kotlin/kotlinx.coroutines/blob/master/CHANGES.md)
- [Commits](https://github.com/Kotlin/kotlinx.coroutines/compare/1.6.1...1.11.0)

---
updated-dependencies:
- dependency-name: org.jetbrains.kotlinx:kotlinx-coroutines-android
  dependency-version: 1.11.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
- dependency-name: org.jetbrains.kotlinx:kotlinx-coroutines-core
  dependency-version: 1.11.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
- dependency-name: org.jetbrains.kotlinx:kotlinx-coroutines-test
  dependency-version: 1.11.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* fix(ktor-client): Send replay network timestamps in milliseconds (#6619)

* fix(ktor-client): Send replay network timestamps in milliseconds

Session Replay reads the HTTP start and end timestamps on a breadcrumb as epoch
milliseconds, which is what sentry-okhttp sends. sentry-ktor-client divided the
epoch nanoseconds by 1000, so it sent microseconds and its replay network spans
landed 1000x too far in the future.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* changelog

* changelog

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(ai): Remove stale sentry-for-ai skills from agents.toml (#6620)

getsentry/sentry-for-ai removed sentry-code-review and
sentry-pr-code-review, dissolved the sentry-workflow router, and moved
sentry-fix-issues to skills-legacy, which dotagents no longer resolves.
Project install on main fails on the first of them.

Also add --project to the dotagents commands in the header. Since
dotagents 3.x the default scope is global (~/.agents/), so the commands
as written do nothing for this repo.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* build(android): Include Android resources in navigation3 unit tests (#6618)

AGP 9.5 registers a Compose Preview task for every Compose module and
fails to create it when unit tests exclude Android resources. Normal
builds never create that task, but anything that creates every task,
such as an LSP's Gradle import or `./gradlew tasks --all`, fails on
navigation3. Every other Android module already sets this.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* chore(deps): bump feign from 11.6 to 13.17 (#6606)

Bumps `feign` from 11.6 to 13.17.

Updates `io.github.openfeign:feign-core` from 11.6 to 13.17
- [Release notes](https://github.com/openfeign/feign/releases)
- [Changelog](https://github.com/OpenFeign/feign/blob/13.17/CHANGELOG.md)
- [Commits](https://github.com/openfeign/feign/compare/11.6...13.17)

Updates `io.github.openfeign:feign-gson` from 11.6 to 13.17
- [Release notes](https://github.com/openfeign/feign/releases)
- [Changelog](https://github.com/OpenFeign/feign/blob/13.17/CHANGELOG.md)
- [Commits](https://github.com/openfeign/feign/compare/11.6...13.17)

---
updated-dependencies:
- dependency-name: io.github.openfeign:feign-core
  dependency-version: '13.17'
  dependency-type: direct:production
  update-type: version-update:semver-major
- dependency-name: io.github.openfeign:feign-gson
  dependency-version: '13.17'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* chore(deps): bump slf4j from 1.7.30 to 2.0.20 (#6536)

* chore(deps): bump slf4j from 1.7.30 to 2.0.20

Bumps `slf4j` from 1.7.30 to 2.0.20.

Updates `org.slf4j:slf4j-api` from 1.7.30 to 2.0.20

Updates `org.slf4j:slf4j-jdk14` from 1.7.30 to 2.0.20

---
updated-dependencies:
- dependency-name: org.slf4j:slf4j-api
  dependency-version: 2.0.20
  dependency-type: direct:production
  update-type: version-update:semver-major
- dependency-name: org.slf4j:slf4j-jdk14
  dependency-version: 2.0.20
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

* Apply suggestion from @adinauer

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Alexander Dinauer <adinauer@users.noreply.github.com>

* chore(deps): bump com.google.devtools.ksp from 2.3.13 to 2.3.14 (#6630)

Bumps [com.google.devtools.ksp](https://github.com/google/ksp) from 2.3.13 to 2.3.14.
- [Release notes](https://github.com/google/ksp/releases)
- [Commits](https://github.com/google/ksp/compare/2.3.13...2.3.14)

---
updated-dependencies:
- dependency-name: com.google.devtools.ksp
  dependency-version: 2.3.14
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* chore(deps): bump androidx.annotation:annotation from 1.9.1 to 1.11.0 (#6627)

Bumps androidx.annotation:annotation from 1.9.1 to 1.11.0.

---
updated-dependencies:
- dependency-name: androidx.annotation:annotation
  dependency-version: 1.11.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* chore(deps): bump org.apache.tomcat:tomcat-catalina from 9.0.108 to 11.0.27 (#6628)

* chore(deps): bump org.apache.tomcat:tomcat-catalina

Bumps org.apache.tomcat:tomcat-catalina from 9.0.108 to 11.0.27.

---
updated-dependencies:
- dependency-name: org.apache.tomcat:tomcat-catalina
  dependency-version: 11.0.27
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>

* Apply suggestion from @runningcode

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: Antonis Lilis <antonis.lilis@sentry.io>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Alexander Dinauer <adinauer@users.noreply.github.com>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Nelson Osacky <nelson.osacky@sentry.io>
Co-authored-by: Adam Brown <adam.brown@sentry.io>
Co-authored-by: Lazizbek Ergashev <20501725+lazerg@users.noreply.github.com>
Co-authored-by: sentry-junior[bot] <264270552+sentry-junior[bot]@users.noreply.github.com>
Co-authored-by: Gene Wright <gene.wright@sentry.io>
Co-authored-by: J-P Nurmi <jpnurmi@gmail.com>
Co-authored-by: Roman Zavarnitsyn <rom4ek93@gmail.com>
Co-authored-by: 0xadam-brown <281682121+0xadam-brown@users.noreply.github.com>
Co-authored-by: sentry-release-bot[bot] <180476844+sentry-release-bot[bot]@users.noreply.github.com>
Co-authored-by: Markus Hintersteiner <markus.hintersteiner@sentry.io>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

sanity-check PR needs a lightweight review for obvious issues

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants