Alec Grieser
Alec Grieser
This adds a new `RecordQueryPlan` that has the same semantics (in terms of results returned, including continuations) as a `RecordQueryIndexPlan`, but that scans additional ranges for the purposes of loading...
This adds `FDBRecordStoreSparseJoinPerformanceTest` for profiling sparse joins. There are some instructions on how to run that test locally within the Javadoc for that test.
At this point, we've been publishing dual proto2 and proto3 versions of the Record Layer for a while. This complicates our builds in a number of ways. We should be...
This makes the `TypeRepository` handling in the `EvaluationContext` API more-buildery, with a default value of `TypeRepository.EMPTY_SCHEMA`. This also simplifies some logic that was creating a child builder off of a...
In PR #1564, the maintained `EvaluationContextBuilder` API was broken by introducing a `build()` method that took a parameter. In order to avoid breaking that API for adopters, we should be...
This changes the single-record delete path so that in cases where we know which key(s) the record used (including taking split points into account), we issue a series of single-key...
At the moment, when a single record is deleted, we (typically) will end up issuing one range delete that covers the whole record. See this implementation in `KeyValueUnsplitter`: https://github.com/FoundationDB/fdb-record-layer/blob/ceaaeb538d3328ad6e3b32f03c34c3764a0195b6/fdb-record-layer-core/src/main/java/com/apple/foundationdb/record/provider/foundationdb/SplitHelper.java#L239-L250 Note...
In Java 14, the JDK introduced "record" as a type of restricted class for "plain data carrying". See: https://docs.oracle.com/en/java/javase/14/language/records.html That seems pretty useful, but unfortunately, it introduces the "record" keyword...
The online indexing API defines a few places where it takes a `Key.Evaluated` (and also usually defines an overload where it takes a `Tuple`). Most of the API, however, uses...
When running tests in CI, I saw the `testIndexFromIndexPersistentContinuePreviousByRecords` test fail with "runner has been closed" errors, though it only happens intermittently. The stack trace is: ``` java.util.concurrent.CompletionException: com.apple.foundationdb.record.provider.foundationdb.FDBDatabaseRunner$RunnerClosed: runner...