Reduce reader planning overhead and qualify against Iceberg - #13

Merged

ethan-tyler merged 18 commits into

mainfrom

perf/reader-planning

Sep 8, 2026

Commits on Sep 7, 2026

feat: authenticate checkpoint pages and add bounded storage ranges
ethan-tyler
committed
167a396
feat: add an authenticated on-demand metadata reader
ethan-tyler
committed
291de2f
feat: integrate pinned OTMP tables with DataFusion 55
ethan-tyler
committed
8c70859
docs: record reader contracts and integrated qualification
ethan-tyler
committed
3e43f28
docs: distinguish Parquet object checks in request counts
ethan-tyler
committed
c72c3ef
test(reader): add file scale and request latency qualification
ethan-tyler
committed
62a8143
fix(qualification): measure process exit without timeout polling
ethan-tyler
committed
44e2a60
test(reader): isolate engine cache capacity at 16k files
ethan-tyler
committed
4135b82
docs(reader): report file-scale qualification and measured limits
ethan-tyler
committed
ca11c04
perf(reader): seek branch membership by indexed file id
ethan-tyler
committed
3bdeacc
perf(reader): batch bounded file metric lookups
ethan-tyler
committed
250286a
perf(reader): use indexed probes for batched metrics
ethan-tyler
committed
ad57768
perf(reader): reuse authenticated pages and immutable revisions
ethan-tyler
committed
0f44992
perf(datafusion): reuse validated footer bytes and qualify planning
ethan-tyler
committed
e49a3c3
fix(reader): validate first-page membership and qualification outcomes
ethan-tyler
committed
83d3e6a
docs(perf): seal reader planning and Iceberg qualification
ethan-tyler
committed
2664cf5
ci: compile and test the independent reader comparator
ethan-tyler
committed
88cefa7

Commits on Sep 8, 2026

Merge landed scale qualification ancestry without changing planning source
ethan-tyler
committed
61b9a29

Navigation Menu

Uh oh!

Reduce reader planning overhead and qualify against Iceberg - #13

Reduce reader planning overhead and qualify against Iceberg#13
ethan-tyler merged 18 commits into
mainfrom
perf/reader-planning

Commits on Sep 7, 2026

feat: authenticate checkpoint pages and add bounded storage ranges

feat: add an authenticated on-demand metadata reader

feat: integrate pinned OTMP tables with DataFusion 55

docs: record reader contracts and integrated qualification

docs: distinguish Parquet object checks in request counts

test(reader): add file scale and request latency qualification

fix(qualification): measure process exit without timeout polling

test(reader): isolate engine cache capacity at 16k files

docs(reader): report file-scale qualification and measured limits

perf(reader): seek branch membership by indexed file id

perf(reader): batch bounded file metric lookups

perf(reader): use indexed probes for batched metrics

perf(reader): reuse authenticated pages and immutable revisions

perf(datafusion): reuse validated footer bytes and qualify planning

fix(reader): validate first-page membership and qualification outcomes

docs(perf): seal reader planning and Iceberg qualification

ci: compile and test the independent reader comparator

Commits on Sep 8, 2026

Merge landed scale qualification ancestry without changing planning source