The target type for the conversion of the complete collection gets adapted when a custom conversion was applied to the elements.
Also `JdbcValue` elements as a result of a custom conversion get unwrapped.
Closes#2078
Original pull request: #2081
TypeInformation.OBJECT must not be compared by reference since TypeInformation instances come form a LRU cache and if TypeInformation.OBJECT gets evicted future instances might be equal, but won't have the same reference.
Closes#2083
If no columns are given, all columns are selected by default.
If columns are specified, only these are selected.
Joins normally triggered by columns from 1:1 relationships are not implemented, and the corresponding columns don't get loaded and can't be specified in a query.
Limiting columns is not supported for single query loading.
Closes#1803
Original pull request: #1967
Deprecate original DialectResolver and JdbcArrayColumns as they've been in the wrong package and introduce replacements in the dialect package. Let deprecated types extend from their replacements to retain compatibility.
Make instance holders final, fix Javadoc typos, update reference docs.
Original pull request #2036
See #2031
On some occasions where col in (:args) contain a really lot args, 10k+ for instance, this commit fixes a performance (high CPU) bug by NOT traversing the whole map in basically O(n^2) manner
Signed-off-by: Mikhail Fedorov <mfedorov761@gmail.com>
Squashed by Jens Schauder
Original pull request #2050
This allows to use it comfortable to configure personal preferences for running maven.
One example is to disable build caches.
Original pull request #2039
Extract SequenceEntityCallbackDelegate from IdGeneratingBeforeSaveCallback. Renameto IdGeneratingEntityCallback and move callback to convert package.
Align return values and associate generated sequence value with the entity. Fix test. Add ticket references to tests.
Extract documentation partials.
See #1955
Original pull request: #2028
Sequence details are now maintained on the property level instead of using the entity level. This is a more accurate representation of the underlying model and that properties are annotated and not entities. It also allows future extension of expanding sequence support to general properties.
Extract delegate for sequence generation. Move types to org.springframework.data.jdbc.core.convert to resolve package cycles.
See #2003
Original pull request: #2005
Refine assignment flow and use early returns where possible. Cache empty MapSqlParameterSource. Reduce dependency on RelationalMappingContext using a lower-level abstraction signature. Simplify names. Use default value check from Commons. Fix log warning message. Add missing since tags.
Remove superfluous annotations and redundant code. Tweak documentation wording.
Closes#2003
Original pull request: #2005