feat(bench): measure the data plane's operation costs and feed them to the Planner (Stage E) by zzylol · Pull Request #835 · ProjectASAP/ASAPQuery-backend

feat!: install the Planner's InstallablePlan; depend on asap-planner and asap-executor only

zzylol

committed

bf9a0f8

chore: ignore Python bytecode caches

zzylol

committed

937bc08

feat(data-plane): sharded ingestion workers, a query thread pool and a shared pane store

zzylol

committed

d32a8bb

fix(data-plane): a query sees the writes before it; a worker closes the watermark it announces

zzylol

committed

1b5d7a9

feat(precompute): update each open pane's summary state per sample

zzylol

committed

1eeac23

feat(data-plane): read query-time raw windows from a local raw buffer and Prometheus remote read

zzylol

committed

cf53170

feat(data-plane): keep each metric in the raw buffer for its own longest query-time read

zzylol

committed

66488dc

feat(data-plane): columnar engine on the Planner's Arrow operators; compile once; per-sample panes from columns

zzylol

committed

19c2570

perf(data-plane): intern series at decode; key every per-series structure by id

zzylol

committed

072bb48

feat(bench): measure the data plane's operation costs and feed them to the Planner

zzylol

committed

3158464

Navigation Menu

Repository navigation

feat(bench): measure the data plane's operation costs and feed them to the Planner (Stage E) - #835

feat(bench): measure the data plane's operation costs and feed them to the Planner (Stage E)#835
zzylol wants to merge 19 commits into
feat/intern-seriesfrom
feat/stage-e-cost-feedback

Commits on Oct 18, 2026

refactor!: plan with ASAPPlanner's stage pipeline and drop the old execution stack

feat(sds): define stored summaries from the plan's ingestion-time outputs

feat(precompute): build stored summary panes from Remote Write samples

feat(query): answer planned instant queries by running the plan's roots

test(e2e): answer #509's PromQL examples like the Planner executor

feat(query): keep query-time panes across evaluations

feat(query): keep a query-time pane only once it is closed

feat(query): answer planned range queries one evaluation per step

test(query): a nonzero series limit is forwarded to Prometheus

Commits on Oct 19, 2026

feat!: install the Planner's InstallablePlan; depend on asap-planner and asap-executor only

chore: ignore Python bytecode caches

feat(data-plane): sharded ingestion workers, a query thread pool and a shared pane store

fix(data-plane): a query sees the writes before it; a worker closes the watermark it announces

feat(precompute): update each open pane's summary state per sample

feat(data-plane): read query-time raw windows from a local raw buffer and Prometheus remote read

feat(data-plane): keep each metric in the raw buffer for its own longest query-time read

feat(data-plane): columnar engine on the Planner's Arrow operators; compile once; per-sample panes from columns

perf(data-plane): intern series at decode; key every per-series structure by id

feat(bench): measure the data plane's operation costs and feed them to the Planner