Port the out-of-core zarr map-reduce engine - #32
Merged
Conversation
Brings the reusable map-reduce ENGINE from vEcoli PR #414 commit 5ec59d33 into
viva-emitters, so any downstream that emits the viva zarr layout can run
out-of-core analyses over it. The concrete moving-average pipeline (Altair/CLI)
stays downstream in the consuming app (v2ecoli).
New modules (viva_emitters/xarray_emitter/):
- mapreduce.py — generic map-reduce algebra (pool_dict/async_dict contexts,
Reducer ABC + ReducerList/Mapping/Distribute, MinMax/Chain/
Unique). Pure stdlib + numpy + bottleneck. Copied verbatim.
- zarr_mapreduce.py — the ZarrMapReduce ABC driver that walks the emitter's
workflow -> substore -> partition -> variable hierarchy,
combining multiprocessing across substores with zarr's async
API within each. Imports rewritten to viva-relative.
- zarr_utils.py — READ-side helpers only (get_async_array/get_async_group,
parse_codecs, get_ndarray, get_rectilinear_ndarray). The
consolidation functions are NOT duplicated here — viva keeps
those in zarr_writer.py. All zarr-internal symbols verified
present on the locked zarr 3.1.6.
storage.py additions: EXPERIMENT/VARIANT/LINEAGE_PREFIX constants; module
functions var_name()/coo_path() + VariablePath; WorkflowConfig / WorkflowPaths /
Substore; and XarrayStoragePartition.{from_substore, independent_path_glob,
get_variant, get_lineage, is_time_coo_name, __hash__}. Reconciled against viva's
diverged partition (which already had independent_path/cast).
Key adaptation: WorkflowConfig.build/.load take an injectable
`variant_parser` (default None) instead of importing
runscripts.create_variants — the library must not depend on app runscripts.
v2ecoli injects its real parser. fsspec is imported lazily (remote stores only).
Deps: bottleneck (+ fsspec) added to the [xarray] extra.
Verification (tests/test_zarr_mapreduce_engine.py): a real viva-emitted
workflow store (1 variant x 2 lineages x 2 generations) is discovered by
WorkflowPaths.locate, XarrayStoragePartition.from_substore round-trips, and
zarr_utils.get_async_group reads the generation=2 array back. Full suite: 110
passed, 4 skipped.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
eagmon
added a commit
that referenced
this pull request
Aug 21, 2026
#32 added bottleneck/fsspec to the [xarray] extra but did not re-lock, and #31 re-locked before those deps existed, so main's uv.lock was missing them (`uv lock --check` failed). Re-lock to restore consistency. Verified: `uv lock --check` passes; full suite 121 passed, 4 skipped. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This was referenced Aug 21, 2026
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Brings the reusable map-reduce engine from vEcoli PR #414 commit
5ec59d33into viva-emitters, so any downstream that emits the viva zarr layout can run out-of-core analyses over it. The concrete moving-average pipeline (Altair/CLI) stays downstream in the consuming app (v2ecoli, PR-D).New modules (
viva_emitters/xarray_emitter/)mapreduce.py— generic map-reduce algebra (pool_dict/async_dictcontexts,ReducerABC +ReducerList/Mapping/Distribute,MinMax/Chain/Unique). Pure stdlib + numpy + bottleneck. Verbatim.zarr_mapreduce.py— theZarrMapReduceABC driver that walks the emitter's workflow → substore → partition → variable hierarchy, combining multiprocessing across substores with zarr's async API within each. Imports rewritten viva-relative.zarr_utils.py— read-side helpers only (get_async_array/get_async_group,parse_codecs,get_ndarray,get_rectilinear_ndarray). The consolidation functions are not duplicated — viva keeps those inzarr_writer.py. All zarr-internal symbols verified present on the locked zarr 3.1.6.storage.pyadditionsEXPERIMENT/VARIANT/LINEAGE_PREFIXconstants; module functionsvar_name()/coo_path()+VariablePath;WorkflowConfig/WorkflowPaths/Substore; andXarrayStoragePartition.{from_substore, independent_path_glob, get_variant, get_lineage, is_time_coo_name, __hash__}. Reconciled against viva's diverged partition (which already hadindependent_path/cast).Key adaptation — no app dependency
WorkflowConfig.build/.loadtake an injectablevariant_parser(defaultNone) instead offrom runscripts.create_variants import parse_variants. The library must not depend on app runscripts; v2ecoli injects its real parser.fsspecis imported lazily (remote stores only).Verification
tests/test_zarr_mapreduce_engine.pybuilds a real viva-emitted workflow store (1 variant × 2 lineages × 2 generations), then assertsWorkflowPaths.locatediscovers both substores,XarrayStoragePartition.from_substoreround-trips, andzarr_utils.get_async_groupreads thegeneration=2array back. Full suite: 110 passed, 4 skipped. (The fullZarrMapReducedriver — ~12 abstract hooks — is exercised by the downstream moving-average pipeline.)Part of the PR #414 reconciliation (follows #29, #30, #31).
🤖 Generated with Claude Code