Fold PhysicalColumnIndexContainer's IndexTypeMap into a presence mask and a dense reader array

[xiangfu0]

What

PhysicalColumnIndexContainer stored its readers in an IndexTypeMap holding a span-sized IndexReader[]: with forward index at numeric id 2 and null vector at 8, a column with two readers allocated seven slots. It now keeps a long presence mask plus a densely packed, exactly sized reader array, so getIndex is a shift, mask and Long.bitCount instead of a range check, at the same O(1) cost. Construction also drops the transient per-column ArrayLists.

A Preconditions.checkState rejects a numeric index id of 64 or above at construction rather than silently misbehaving; OSS uses 13 ids today.

Tests

PhysicalColumnIndexContainerTest: forward only, forward plus null vector, and a mix spanning the lowest and a high id, asserting getIndex returns the created reader for present types and null for absent ones, that forwardIndexOnly filtering still applies, and that close closes each reader exactly once.

Why

A server keeps one metadata object graph per (segment, column) for as long as the segment is loaded, so on wide tables the per-column footprint decides how many segments a server can hold. Measured end to end on a 1000-column segment, this series takes the heap retained at load from 4.08 MB to 0.175 MB per segment (4,080 to 174 bytes per column), with a fully compacting collector on both sides. No on-disk format change, the /tables/{table}/segments/{segment}/metadata JSON stays byte-identical, and every public and SPI signature keeps working.

Stack

Part 4 of 9, based on #19474. Review only this part's own commits; the earlier parts account for the rest of the diff.

1. Allocate ColumnMetadataImpl index sizes lazily and skip the index_map lookup without an index dir #19480 lazy index-size storage
2. Canonicalize the default null value and intern per-column strings at metadata parse time #19473 canonical default-null values and interned per-column strings
3. Delegate immutable DataSourceMetadata to ColumnMetadata instead of snapshotting it #19474 delegating immutable DataSourceMetadata
4. Fold PhysicalColumnIndexContainer's IndexTypeMap into a presence mask and a dense reader array #19475 presence-mask index container
5. Share segment-derived FieldSpec instances across segments through a weak interner #19476 weak FieldSpec interner
6. Materialize immutable-segment columns lazily behind an opt-in instance config (default off) #19477 opt-in lazy column materialization
7. Slim ColumnMetadataImpl to 72 bytes and derive the per-segment Schema lazily #19478 slim ColumnMetadataImpl and lazy per-segment Schema
8. Store numeric column min/max as primitives instead of boxed Comparables #19479 primitive numeric min/max
9. Hold segment column metadata in sorted arrays and derive the map on demand #19481 sorted-array column metadata store

[codecov-commenter]

Codecov Report

❌ Patch coverage is 72.72727% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 57.80%. Comparing base (2b73f0d) to head (0bbde43).

Files with missing lines	Patch %	Lines
...ent/index/column/PhysicalColumnIndexContainer.java	72.72%	7 Missing and 2 partials ⚠️

Additional details and impacted files

@@                              Coverage Diff                              @@
##             xiangfu0/data-3221-3-datasource-adapter   #19475      +/-   ##
=============================================================================
- Coverage                                      57.81%   57.80%   -0.02%     
- Complexity                                         1        7       +6     
=============================================================================
  Files                                           2688     2688              
  Lines                                         164324   164308      -16     
  Branches                                       26683    26681       -2     
=============================================================================
- Hits                                           95008    94977      -31     
- Misses                                         61291    61293       +2     
- Partials                                        8025     8038      +13     

Flag	Coverage Δ
integration	100.00% <ø> (+100.00%)	⬆️
integration1	100.00% <ø> (?)
integration2	0.00% <ø> (ø)
java-25	57.80% <72.72%> (-0.02%)	⬇️
lane-a	100.00% <ø> (+100.00%)	⬆️
lane-b	0.00% <ø> (ø)
temurin	57.80% <72.72%> (-0.02%)	⬇️
unittests	57.80% <72.72%> (-0.02%)	⬇️
unittests1	57.80% <72.72%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.