Master->Dev Update #98
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
PiperOrigin-RevId: 316809187
…lay mode. FutureProxyManager still grabs FutureHandle that supposed to be cleaned up and released. Force release them during proxy manager's destructor. PiperOrigin-RevId: 316809477
PiperOrigin-RevId: 317136423
…m 6.26.0 PiperOrigin-RevId: 317141369
PiperOrigin-RevId: 317141369
PiperOrigin-RevId: 317141369
Add empty workflow to manually trigger.
Also fixes a flaky test because assertion is some times done before expected remote event arrives. PiperOrigin-RevId: 317154095
…g predicates.
See b/158478280 and go/absl-cleanup-lsc for more details.
This change was produced using rename_function with the spec:
rename {
rename_spec {
new_header: "third_party/absl/strings/str_split.h"
old_name: "strings::Split"
new_name: "absl::StrSplit"
}
}
Additionally, a global find/replace was done on:
strings::AllowEmpty -> absl::AllowEmpty
strings::SkipEmpty -> absl::SkipEmpty
strings::SkipWhitespace -> absl::SkipWhitespace
Tested:
TAP sample presubmit queue
http://test/OCL:317011231:BASE:317041434:1592460837225:fc71f604
PiperOrigin-RevId: 317399788
PiperOrigin-RevId: 317402209
*** Reason for rollback *** Not ready for release yet *** Original change description *** Add a user callback executor for android. Also fixes a flaky test because assertion is some times done before expected remote event arrives. *** PiperOrigin-RevId: 317736475
*** Reason for rollback *** Not ready for release yet *** Original change description *** [C++] Provide a default executor with settings on iOS *** PiperOrigin-RevId: 317791137
Also fix a memory leak during listener unregistration PiperOrigin-RevId: 318538053
Fix leaks in firestore_test that the HeapChecker found. Test with `blaze test -c dbg` to get line numbers in stack frames. PiperOrigin-RevId: 318644593
PiperOrigin-RevId: 318885157
*** Reason for rollback *** re-roll cl/317154095 SKIP_FIRESTORE_KOKORO_BUILD_TEST=true *** Original change description *** Automated g4 rollback of changelist 317154095. *** Reason for rollback *** Not ready for release yet *** Original change description *** Add a user callback executor for android. Also fixes a flaky test because assertion is some times done before expected remote event arrives. *** *** PiperOrigin-RevId: 318891491
PiperOrigin-RevId: 318899934
*** Reason for rollback *** Re-roll cl/314233757 SKIP_FIRESTORE_KOKORO_BUILD_TEST=true *** Original change description *** Automated g4 rollback of changelist 314233757. *** Reason for rollback *** Not ready for release yet *** Original change description *** [C++] Provide a default executor with settings on iOS *** *** PiperOrigin-RevId: 318938740
There are several related flaws fixed here:
* The mutex in TestEventListener was not applied consistently, leading to
data races that showed up as strangely failing tests on forge.
* The result of FirestoreIntegrationTest::Await was implicitly trusted to
produce some results, but this isn't the case when it times out. In this
error case the caller would read uninitialized memory almost immediately
after, leading to crashes before the log message about the timeout had
necessarily been writen.
* The result of FirestoreIntegrationTest::Await was implicitly trusted not to
produce more results than requested, but this didn't always happen either.
This would cause failures where a test would request `n` events, and use
the last `n` that arrived, skipping extra events. Now the EventAccumulator
ensures that it consumes events in sequence.
PiperOrigin-RevId: 318943435
Certain tests include a warm-up step to ensure that the backend was actually available, but only waited for any event. Unfortunately this doesn't work, because when the server is unavailable, the SDK will serve a from-cache event indicating the document doesn't exist. Change all these to wait for a from-cache: false event, guaranteeing that the server is actually available and confirming the document doesn't exist. PiperOrigin-RevId: 318948824
This simplifies the specification of promises, cleanup functions, and converters, since they now only need to specify the public type. PiperOrigin-RevId: 318960642
This makes promise creation more closely match iOS and paves the way for removing WrapperFuture altogether. Use PromiseFactory in all internal type implementations instead of extending WrapperFuture and then dDelete WrapperFuture. PiperOrigin-RevId: 319025783
PiperOrigin-RevId: 319034882
Previously, setting FirebaseFirestore.LogLevel to LogLevel.Debug would get bumped back to LogLevel.Info as an unexpected side effect of creating of a new firebase::App object. The apparent effect to users was that enabling debug logging had no effect. This made it challenging to debug issues with customers because they were unable to collect valuable debug logs when using iOS as their platform. PiperOrigin-RevId: 319056991
This CL is the last in a series described in cl/312713181, cl/317397413, and resolves b/156024690. In this CL the new pattern for managing listener callbacks implemented in `FirebaseFirestore.ListenForSnapshotsInSync()` is being applied to all other places where we expose listeners (i.e. `Query` and `DocumentReference`). PiperOrigin-RevId: 319089024
THIS CHANGE IS BELIEVED SAFE
Templated asynchronous code can change the lifetime of string data as a result
of this change; however, the most common uses of these (lambdas and callbacks)
are excluded from this change.
Further, your TAP tests pass.
go/string-ref-to-string-view-lsc
Tested:
TAP --sample ran all affected tests and none failed
http://test/OCL:319219698:BASE:319213711:1593616691507:a583f044
PiperOrigin-RevId: 319286243
This has been observed during cleanup when a DocumentReferenceInternal is destroyed, its Future API can end up deleting orphaned Future APIs that contain Futures holding the containing `DocumentReference`. PiperOrigin-RevId: 320260380
Wrapping macro bodies in do { } while(false) makes them into a statement that
legitimately should be followed by a semicolon.
PiperOrigin-RevId: 320272779
* Avoid dereferencing awaited pointers after Await has failed, preventing
crashes after test timeouts.
* Await `CollectionReference::Add`, preventing nondeterminism in tests.
PiperOrigin-RevId: 320449430
This was achieved by using the CppInstanceManager, which provides this functionality elsewhere in the Unity SDK. When running UIHandlerAutomated, this eliminates all 11 leaks from Firestore::GetFirestore(). PiperOrigin-RevId: 320609048
These calls were present as a workaround for bugs in the Firestore destructor; however, since those bugs have been fixed the calls to Terminate() are now superfluous and, worse, can hide bugs. As a result, they are being removed. PiperOrigin-RevId: 320627333
This is causing Rapid build fail with error message like "error: no template named 'make_unique' in namespace 'std'; did you mean 'MakeUnique'?" Since we have our own implementation, we should use it consistently. PiperOrigin-RevId: 320681249
PiperOrigin-RevId: 320688041
Previously, if the value was of the wrong type, you'd get a message like:
Value of: snap.Get("sum").is_integer()
Actual: false
Expected: true
Which would give no indication of what the actual type or value was. Now tests
will fail like this:
Expected equality of these values:
snap.Get("sum")
Which is: 1337 (Type::kDouble)
FieldValue::Integer(value)
Which is: 1337 (Type::kInteger)
As an added bonus this also simplifies the calling code because now we can just
assert that a value in a snapshot is equal to some expected value and
GoogleTest will do the heavy lifting of printing the differences.
One unsolved problem with this approach is how to handle equality within
epsilon for floating point values. This turns out to be non-trivial without
writing custom matchers, which is beyond the scope of what I wanted to tackle
here. Instead of solving this I've changed the tests to use values that have an
exact representation as doubles. This is easier to do for integral values than
for fractional ones so the tests now use integer-valued doubles to achieve the
same effect of cumulative addition as before.
PiperOrigin-RevId: 321022746
This is the first in a series of commits that aims to convert our JNI usage to a more modern approach while still remaining STLPort compatible. PiperOrigin-RevId: 321059370
These generate local and global reference subtypes of a given JNI wrapper that make it possible to automatically emit DeleteLocalRef and DeleteGlobalRef calls in the course of regular usage. PiperOrigin-RevId: 321175591
PiperOrigin-RevId: 321183165
The generated headers are currently missing a newline prior to @ifdef cpp_examples which is causing doc generation to fail. PiperOrigin-RevId: 321272103
PiperOrigin-RevId: 321475293
Previously, all types needed to be in the JniTypeMap, but this was never intended to be the end state because it would require an entry for all subtypes of Object which couldn't scale. The new implementation based on ranked choice overload selection also resolves ambiguities. For example: unsigned char could be a uint8_t (which converts to jbyte) or the underlying type for jboolean. Absent the ranking, an argument of type unsigned char would resolve to multiple overloads and would be ambiguous. PiperOrigin-RevId: 321835773
PiperOrigin-RevId: 322398533
|
We found a Contributor License Agreement for you (the sender of this pull request), but were unable to find agreements for all the commit author(s) or Co-authors. If you authored these, maybe you used a different email address in the git commits than was used to sign the CLA (login here to double check)? If these were authored by someone else, then they will need to sign a CLA as well, and confirm that they're okay with these being contributed to Google. ℹ️ Googlers: Go here for more info. |
|
A Googler has manually verified that the CLAs look good. (Googler, please make sure the reason for overriding the CLA status is clearly documented in these comments.) ℹ️ Googlers: Go here for more info. |
DellaBitta
changed the title
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
7 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.