* Revert "Revert "Skip filterIter match check when a key range is contiguous (#8242)""
This reverts commit 277c1d7d8b.
* fix string matching bug, test added in GMS#2639
* Better comment
* Skip filterIter match check when a key range is contiguous
* [skip actions] [ga-update-correctness] SQL Correctness updated to 99.960017
* certain encodings are imperfectly serialized, still need range.Matches check
* better comments
* inline contigous check
* [skip actions] [ga-update-correctness] SQL Correctness updated to 100
* bad default value
* [skip actions] [ga-update-correctness] SQL Correctness updated to 99.991949
* fix nil test (added in GMS 2630)
* [skip actions] [ga-update-correctness] SQL Correctness updated to 100
* zach comments
---------
Co-authored-by: coffeegoddd <coffeegoddd@users.noreply.github.com>
* add three way differ for merge optimizations
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* docstrings
* prog
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* fix tests
* format
* more test fixes
* rewrite chunk of table merge, still need to rewrite unique lookup checking
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* more test fixes
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* test
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* fix tests
* batching test
* larger buffer for secondary mutableMap
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* refactor, fix buffer sizing
* fix more tests
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* restore proto
* sort unique checks by secondary
* [ga-format-pr] Run go/utils/repofmt/format_repo.sh and go/Godeps/update.sh
* fix keyless pk len
* fmt
* revert edits
* merge conflicts
* merge conflict
---------
Co-authored-by: max-hoffman <max-hoffman@users.noreply.github.com>
At one point, we started to trim null suffixes from tuples. In versions prior to this change,
tuples could have a null field at the end of the tuple. Now we always truncate the tuple.
If old versions insert rows with null values as the last column, it produces a tuple
that is longer than the tuple that would be inserted by the latest version. This leads to
empty diffs.
Maximilian Hoffman