Optional Compression

[benjaminwinger]

Not quite done since it's currently only exposed in the C++ and Rust APIs, and the option needs to be passed to NodeColumns.

Python and C don't currently have any sort of SystemConfig structure and only take the buffer pool size as an argument. (Edit: Added for python and C APIs)
It looks the same for nodejs and java (which I'm much less familiar with).

An optional argument would probably work fine for python, but since C lacks default values it should probably also have a SystemConfig struct that can be passed to kuzu_database_init. (done)

[codecov]

Codecov Report

Attention: 3 lines in your changes are missing coverage. Please review.

Comparison is base (f1f7b75) 89.63% compared to head (5164b3e) 89.66%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2102      +/-   ##
==========================================
+ Coverage   89.63%   89.66%   +0.03%     
==========================================
  Files         988      988              
  Lines       35717    35729      +12     
==========================================
+ Hits        32014    32036      +22     
+ Misses       3703     3693      -10     

Files	Coverage Δ
src/c_api/database.cpp	85.00% <100.00%> (+1.66%)	⬆️
.../include/processor/operator/persistent/copy_node.h	100.00% <100.00%> (ø)
src/include/storage/local_storage.h	100.00% <ø> (ø)
src/include/storage/local_table.h	100.00% <100.00%> (ø)
src/include/storage/storage_manager.h	100.00% <100.00%> (ø)
src/include/storage/store/column_chunk.h	90.74% <100.00%> (+0.17%)	⬆️
src/include/storage/store/node_column.h	100.00% <ø> (ø)
src/include/storage/store/node_group.h	100.00% <ø> (ø)
src/include/storage/store/node_table.h	100.00% <ø> (ø)
src/include/storage/store/nodes_store.h	100.00% <100.00%> (ø)
... and 22 more

... and 2 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ray6080]

Is there a plan to test the configuration inside cpp?

[benjaminwinger]

Is there a plan to test the configuration inside cpp?

Not really. We could maybe expose information about the properties which could include the compression information for each column chunk. That could be useful later for examining the efficiency of the compression (e.g. if it also reports the compressed size and the number of values stored).

[ray6080]

We should also add an arg nocompression to CLI.