Skip to content

Latest commit

 

History

History
executable file
·
177 lines (123 loc) · 5.77 KB

CHANGELOG.md

File metadata and controls

executable file
·
177 lines (123 loc) · 5.77 KB

Changelog

Version 2.0.2

2023-03-09

Fix embedded proguard rules to use -keepnames instead of -keeppackagenames as the latter isn't respected by R8 3+ when repackageclasses is enabled.

Thanks to @viakunin for contributing this fix!

Version 1.0.2

2023-03-09

Same proguard changes as 2.0.2 but for RxJava 2.x/RxDogTag 1.x.

Version 2.0.1

2020-10-26

Fix embedded proguard rules to completely keep subscribe() methods and rxdogtag code. R8 may inline subscribe() calls entirely to their call-sites, which breaks RxDogTag's tagging.

While this blanket keep is sort of a cardinal sin of libraries to do this, RxDogTag is only a few classes and incredibly small, so we claim this is a small price to pay for actionable stacktraces.

Version 1.0.1

2020-10-26

Same proguard changes as 2.0.1 but for RxJava 2.x/RxDogTag 1.x.

Version 2.0.0

2020-03-08

RxDogTag 2 is built against RxJava 3 and is binary-compatible with RxDogTag 1.x and RxJava 2.x. As such - it has a different package name and maven group ID.

Packaging

Maven Group ID Package Name
1.x com.uber.rxdogtag com.uber.rxdogtag
2.x com.uber.rxdogtag2 rxdogtag2

For any sub-packages, the above mapping should be used for those package prefix replacements as well.

Changes

  • Dependency Upgrades
RxJava 3.0.0
AutoDispose 2.0.0

Version 1.0.0

2020-03-07

  • Stable release!
  • This release is identical in functionality to 0.3.0.
  • Various dependency upgrades. (#55)
RxJava 2.2.18
AutoDispose 1.4.0

Version 0.3.0

2019-09-23

New: Builder option to disable guarded observer callbacks. #43

By default, RxDogTag will try/catch every onNext/onSuccess/onComplete calls in its observers to try to catch exceptions and modify the trace before they're handed to the (unhandled) onError. If you only want to handle upstream errors, you can disable this behavior in the builder now. This can be useful in hot paths where performance is more of a concern.

RxDogTag.builder()
    .guardObserverCallbacks(false)
    .install();

New: ErrorReceivers API. #45

For users that have custom onError() implementations but want to supplement it with RxDogTag, there are two new APIs to help with this.

RxDogTagErrorReceiver is a simple interface that your custom observer can implement to indicate that it wants its onError() method to be called with the aforementioned guarded observer callback. Note however that modified tagged exceptions in this case will be delivered to RxJavaPlugins.onError().

RxDogTagTaggedExceptionReceiver is an interface that your custom observer can implement to indicate that it wants its onError() called with the modified tagged exception. This is useful if you have custom onError() handling but want to use/report RxDogTag's tagged information for something custom.

Misc

Special thanks to @emartynov for contributing to this release!

Version 0.2.0

2019-05-14

New: Builder-based API for configuration

RxDogTag now uses a builder API for configuration. Currently this includes the existing ObserverHandler and package whitelisting, as well as a new configuration to optionally disable stacktrace annotations.

RxDogTag.builder()
    .disableAnnotations()
    .addObserverHandlers(...)
    .addIgnoredPackages(...)
    .configureWith(...) // For other custom AutoDispose.Configurers
    .install();

Note: No-config RxDogTag.install() still exists, but is now just a proxy to RxDogTag.builder().install()

Breaking changes:

  • ObserverHandler no longer handles package whitelisting, as this is now moved to the builder API.
  • The AutoDispose artifact now uses the builder API's configureWith support, and AutoDisposeObserverHandler is now just a package-private implementation detail.
RxDogTag.builder()
    .configureWith(AutoDisposeConfigurer::configure)
    .install();

New: Inferred subscribe point is now the first element in the stacktrace.

After discussions with others in the community, the inferred subscribe point is now the first element in the stacktrace for better grouping with crash reporters (which may have otherwise accidentally grouped the stacktrace header annotations as the "same" crash). This plus some arrow characters means the stacktraces look like this:

io.reactivex.exceptions.OnErrorNotImplementedException: The mapper function returned a null value.

Caused by: java.lang.NullPointerException: The mapper function returned a null value.
	at com.uber.anotherpackage.ReadMeExample.complex(ReadMeExample.java:55)
	at [[ ↑↑ Inferred subscribe point ↑↑ ]].(:0)
	at [[ ↓↓ Original trace ↓↓ ]].(:0)
	at io.reactivex.internal.functions.ObjectHelper.requireNonNull(ObjectHelper.java:39)
	at io.reactivex.internal.operators.observable.ObservableMap$MapObserver.onNext(ObservableMap.java:57)
	at io.reactivex.internal.operators.observable.ObservableSubscribeOn$SubscribeOnObserver.onNext(ObservableSubscribeOn.java:58)
	at io.reactivex.internal.operators.observable.ObservableScalarXMap$ScalarDisposable.run(ObservableScalarXMap.java:248)
	// ... and so on

Version 0.1.0

2019-04-09

Initial release!