Spool index backend

[d-chambers]

Replaces DASCore's PyTables/HDF5 spool index with a relational SQLite metadata engine and unifies every spool type (in-memory, directory, and their unions) on top of it. Selection, candidate identification, chunk planning, and union now go through one lazy PatchCatalog, so the same pushdown-capable semantics apply whether patches come from files or memory. The old ChunkManager and PyTables index are removed.

Motivation

The old index stored a flat, per-directory PyTables table and only served directory spools; in-memory and union spools each had their own ad-hoc metadata paths, and chunking went through a separate ChunkManager. Behavior diverged across spool types, queries kept everything in memory, and scaling to large archives was awkward. A relational index with a real query planner lets one code path answer counts, selections, and chunk plans lazily and push work down to the database.

What changed

• Relational backend (io/index/). A normalized seven-table schema (sources, patches, attrs, coordinate definitions, links) with keys, value-kind checks, and cascade foreign keys, validated before any mutation. Coordinate definitions separate an exact fingerprint from a summary-only dedup key. Nullable nanosecond columns are fetched without a float64 round trip (ns epochs exceed 2**53). Ingest normalizes unit-bearing values to canonical base SI, and the query builder compiles the selector spec to SQL (attrs exact, regex as candidate + residual, coordinate predicates candidacy-only with exact re-trim at load).
• One engine for every spool. PatchCatalog owns the backend, a resolver, and the directory synchronizer; directory and memory spools share it, so lazy selection, pushdown, and chunk planning are identical across types. spool + spool unions catalogs table-to-table. In-memory patches carry an eager instance identity, giving spools set semantics by patch identity.
• Chunk planning. ChunkManager is replaced by an inspectable ChunkPlan, exposed publicly as Spool.chunk_plan(...) with the same arguments as chunk.
• Unit-aware selection. Numeric coordinate summaries are stored in canonical SI, so a bare range means SI regardless of native units and the exact per-patch trim defers its representation until each patch is known — correct across mixed unitful/unitless archives. Coordinate selection accepts a (start, stop) range or a boolean mask; scalar/membership selectors are rejected eagerly.
• Relocations/removals. PyTables dropped from pyproject.toml/environment.yml; the chunk planner, path classifiers, and relative-selection helpers moved to utils/.

Summary by CodeRabbit

• New Features

  • Added SQLite-backed directory indexing with incremental updates and a catalog-based patch selection workflow.
  • Expanded spool functionality: richer select options (_attrs/_coords, samples, relative), chunk_plan, and chunk(..., group, missing_dim), plus spool-union support (spool + spool).
  • Added config controls for chunk grouping (sampling_group_tolerance, groupby_attrs).

• Bug Fixes

  • Improved patch identity/serialization behavior and corrected selection/trim exactness, including unit-aware and regex-chaining query handling.
  • Refined chunk/merge validation and edge-case behavior.

• Documentation

  • Updated spool index/selection/chunking docs and directory indexer tutorial for the SQLite model.

• Chores

  • Migrated HDF5 handling/tests from PyTables to h5py; updated related configs and fixtures.

Changelog

• changed breaking: the spool hierarchy collapses to one concrete dc.Spool; MemorySpool, DirectorySpool, FileSpool, the dascore.clients package, and the select_kwargs parameter are removed — use dc.spool(...) and spool.has_live_patches.
• changed breaking: Spool.update() is allowed only on a root spool; any derived spool (select, slice, sort, chunk, +) raises.
• changed breaking: Spool.select no longer accepts coordinate boolean masks — use spool.map(lambda p: p.select(dim=mask)); spool[bool_array] still selects membership and samples=True ranges still apply per patch.
• changed breaking: directory indexes use a seven-table SQLite schema in .dascore_index.sqlite3; the PyTables index is gone, so existing indexes must be deleted and rebuilt.
• changed breaking: SQLite supports concurrent readers and one serialized writer on local filesystems; network filesystems with unreliable locking are not supported.
• removed breaking: PyTablesReader, PyTablesWriter, and the HDF5Reader/HDF5Writer aliases from dascore.utils.hdf5; use H5Reader/H5Writer.
• changed: spools present patches in a defined order — construction order for patch lists, time order for directories, and rows without a value for the ordering key sort last — and equality compares effective contents rather than backing.
• changed: a + b bakes each operand's pending trims and sort order into the union instead of dropping them, including per-patch time order for interleaved multi-patch files.
• changed: Spool.sort(...) accepts any coordinate of the spool, not just time/distance, and contiguous descending-coordinate patches now merge under chunk(dim=None).
• changed: chunking is defined on dimensions only and is unit-aware — compatible spellings such as metres and feet merge, incompatible dimensionalities can never plan together — and chunking a restructured spool along a different dimension plans over its current patches, while re-chunking the same dimension still re-plans from the original members.
• changed: Spool.select validates unknown names and quantity dimensionality, composes chained regexes with AND, supports explicit _attrs/_coords namespaces, and skips (with a warning) attribute values whose units are dimensionally incompatible across files instead of failing the index update.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

📝 Walkthrough

Walkthrough

The PR replaces legacy directory indexing with SQLite-backed catalogs, adds lazy spool selection, union, and chunk planning, normalizes patch identities, and migrates HDF5 utilities and tests from PyTables to h5py.

Changes

Indexing, catalogs, and directory spools

Layer / File(s)	Summary
SQLite index and directory synchronization dascore/io/index/*, dascore/io/indexer.py	Adds a versioned SQLite schema, normalized ingestion records, typed queries, transactional source updates, incremental directory scanning, and catalog-based patch resolution.
DirectorySpool integration dascore/clients/dirspool.py, dascore/core/summary.py, dascore/io/core.py	Routes directory spools through PatchCatalog, normalizes source patch IDs, and resolves loaded patches through catalog resolvers with exact trimming.
Catalog-native spool operations dascore/core/spool.py, dascore/utils/pd.py	Adds lazy selection, _attrs/_coords namespaces, relative and sample-local selection, catalog unions, patch identity semantics, and materialized-spool transitions.
Metadata-only chunk planning dascore/utils/chunk_plan.py, dascore/utils/chunk.py, dascore/utils/coordmanager.py, dascore/utils/patch.py, dascore/config.py	Replaces ChunkManager with deterministic ChunkPlan generation, grouping and tolerance controls, missing-dimension policies, conflict handling, and updated coordinate merge assembly.
HDF5 migration and supporting updates dascore/utils/hdf5.py, dascore/io/h5simple/utils.py, environment.yml, pyproject.toml	Replaces PyTables-oriented HDF5 handling and dependencies with managed h5py resources, including local and remote read/write paths.
Tests and documentation tests/**, docs/**, scripts/_templates/_quarto.yml	Adds coverage and documentation for indexing, selection, chunk planning, unions, identity, incremental updates, and the h5py migration.

Possibly related PRs

• DASDAE/dascore#532 — overlaps in coordinate-merge conflict handling and merge-kwargs routing.
• DASDAE/dascore#584 — overlaps in spool iteration behavior for differing coordinate ranges.
• DASDAE/dascore#743 — overlaps in DirectorySpool patch-loading and resolver behavior.

Suggested labels: spool, IO

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 inconclusive)

Check name	Status	Explanation	Resolution
Title check	❓ Inconclusive	The title is related to the main change, but it is broad and generic rather than specific.	Use a more specific title such as "Replace PyTables spool index with SQLite PatchCatalog backend".

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description check	✅ Passed	The description covers the problem, motivation, major changes, and removals, so it is mostly complete.
Docstring Coverage	✅ Passed	Docstring coverage is 91.35% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch spool-index-backend

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (a79e8e5) to head (e9a873c).

Additional details and impacted files

@@             Coverage Diff             @@
##               dev      #751     +/-   ##
===========================================
  Coverage   100.00%   100.00%             
===========================================
  Files          152       161      +9     
  Lines        14929     16619   +1690     
===========================================
+ Hits         14929     16619   +1690     

Flag	Coverage Δ
network	49.66% <41.93%> (-1.97%)	⬇️
unittests	99.98% <100.00%> (+<0.01%)	⬆️

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:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[coderabbitai]

Actionable comments posted: 8

🧹 Nitpick comments (5)

dascore/io/index/query.py (1)

216-216: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Add strict=True to zip() for safety.

If rows["value_kind"] and rows["column_name"] differ in length, zip() silently truncates. Using strict=True catches this at runtime.

♻️ Proposed fix

-    columns = dict(zip(rows["value_kind"], rows["column_name"]))
+    columns = dict(zip(rows["value_kind"], rows["column_name"], strict=True))

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@dascore/io/index/query.py` at line 216, Update the zip call that builds
columns in the query logic to pass strict=True, ensuring mismatched value_kind
and column_name lengths raise an error instead of being silently truncated.

Source: Linters/SAST tools

dascore/core/patch.py (1)

103-105: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Minor: declare _instance_id alongside other private fields.

The other private/public fields (_data, data, coords, etc.) are declared as class-level annotations; _instance_id isn't, which makes it a bit harder to discover as part of the object's persistent state.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@dascore/core/patch.py` around lines 103 - 105, Declare the _instance_id field
as a class-level annotation alongside the existing _data, data, and coords
fields, while retaining its eager uuid4().hex assignment in the initializer.

dascore/io/index/ingest.py (2)

233-239: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Add stacklevel to the reserved-attr-name warning.

Without stacklevel=2, the warning's reported source location points into ingest.py instead of the caller's ingest call, making it harder to trace which write triggered it.

♻️ Suggested fix

-            warnings.warn(msg, UserWarning)
+            warnings.warn(msg, UserWarning, stacklevel=2)

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@dascore/io/index/ingest.py` around lines 233 - 239, Update the warnings.warn
call in the reserved-attribute handling block to pass stacklevel=2, so the
warning points to the caller of the ingest operation rather than ingest.py.

Source: Linters/SAST tools

---

211-216: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Bare except Exception: pass swallows all errors in the datetime fallback.

This is presumably intentional (classify-or-skip for unclassifiable attrs), but catching blind Exception can mask real bugs (e.g., a broken to_datetime64 call) instead of just skipping non-datetime-like values.

♻️ Suggested narrowing

     try:
         return TypedValue("time", to_int(to_datetime64(value)))
-    except Exception:
+    except (TypeError, ValueError, OverflowError):
         pass

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@dascore/io/index/ingest.py` around lines 211 - 216, Narrow the exception
handling around the datetime fallback in the ingest classification flow to catch
only the expected conversion/parsing errors from to_datetime64 and to_int.
Preserve returning the time TypedValue for valid datetime-like inputs and
returning None for unsupported values, while allowing unexpected programming or
runtime errors to propagate.

Source: Linters/SAST tools

dascore/io/index/indexer.py (1)

32-35: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Consider sharing the hidden-columns constant with catalog.py.

_SPOOL_HIDDEN_COLUMNS here duplicates the inline list in catalog.py's to_df() (["n_dims", "sample_count_total", "shape"]). Extracting a single shared constant would prevent silent drift if one site changes without the other.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@dascore/io/index/indexer.py` around lines 32 - 35, Share the hidden-column
definition between the indexer and catalog code instead of maintaining duplicate
values. Extract or reuse a single module-level constant for "n_dims",
"sample_count_total", and "shape", update _SPOOL_HIDDEN_COLUMNS and catalog.py's
to_df() to reference it, and preserve the existing filtering behavior.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@dascore/core/spool.py`:
- Around line 807-834: Update _as_catalog_member and _chunk_working_df to
preserve DirectorySpool constructor select_kwargs: retain the selection in the
catalog view or apply the equivalent filter before returning catalog membership
and before constructing chunk-planner input. Ensure catalog-native unions and
chunking cannot reintroduce rows excluded by the constructor selection, while
preserving existing behavior for unrestricted spools.
- Around line 1330-1342: Update __rich__ to guard against a missing dataframe
before calling len(df) or accessing df.columns. Preserve the existing time-span
rendering for non-empty dataframes containing time_min, while allowing
MemorySpool() with _df set to None to return the base representation without
error.

In `@dascore/io/index/indexer.py`:
- Around line 92-125: Update _find_index_path so the explicit index_path branch
returns the same absolute Path value recorded in the index map, rather than the
original potentially relative Path(index_path). Preserve the existing map update
behavior and all other path-selection branches.

In `@dascore/utils/patch.py`:
- Around line 630-634: Update the field construction near coord_fields in the
patch formatting logic to flatten all zipped min/max coordinate pairs into a
single list before concatenating with attrs. Replace the invalid multi-argument
list conversion while preserving the existing field order and reindex behavior
for single- and multi-coordinate inputs.

In `@dascore/utils/paths.py`:
- Around line 33-44: The directory_writable() helper should return False for any
filesystem OSError, including read-only mount errors. Move the
probe.parent.mkdir, file creation, and cleanup operation inside the try block,
catch OSError, and preserve the existing successful cleanup and True return
behavior.

In `@dascore/utils/pd.py`:
- Around line 57-59: Update the validation guard in the relative-select logic to
require both envelope columns, lo_col and the corresponding upper-bound column,
before either is accessed. Preserve the existing InvalidSpoolQueryError path for
missing columns or empty data so incomplete metadata cannot reach the query and
raise KeyError.

In `@tests/test_core/test_patch_chunk.py`:
- Around line 309-316: Strengthen test_merge_unequal_other by comparing the
sorted distance_min/distance_max envelopes of distance_adjacent with those in
out, in addition to the existing length assertion. Verify that each unequal
coordinate is preserved exactly and outputs are not duplicated or altered.

In `@tests/test_core/test_spool.py`:
- Around line 968-974: Update test_repr_without_time_coordinate to match
MemorySpool.__rich__: since the patch has no time_min coordinate, assert that
the rendered output does not contain “Time Span,” unless the renderer is
explicitly changed to add a no-time placeholder.

---

Nitpick comments:
In `@dascore/core/patch.py`:
- Around line 103-105: Declare the _instance_id field as a class-level
annotation alongside the existing _data, data, and coords fields, while
retaining its eager uuid4().hex assignment in the initializer.

In `@dascore/io/index/indexer.py`:
- Around line 32-35: Share the hidden-column definition between the indexer and
catalog code instead of maintaining duplicate values. Extract or reuse a single
module-level constant for "n_dims", "sample_count_total", and "shape", update
_SPOOL_HIDDEN_COLUMNS and catalog.py's to_df() to reference it, and preserve the
existing filtering behavior.

In `@dascore/io/index/ingest.py`:
- Around line 233-239: Update the warnings.warn call in the reserved-attribute
handling block to pass stacklevel=2, so the warning points to the caller of the
ingest operation rather than ingest.py.
- Around line 211-216: Narrow the exception handling around the datetime
fallback in the ingest classification flow to catch only the expected
conversion/parsing errors from to_datetime64 and to_int. Preserve returning the
time TypedValue for valid datetime-like inputs and returning None for
unsupported values, while allowing unexpected programming or runtime errors to
propagate.

In `@dascore/io/index/query.py`:
- Line 216: Update the zip call that builds columns in the query logic to pass
strict=True, ensuring mismatched value_kind and column_name lengths raise an
error instead of being silently truncated.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: d2796497-c7c1-41a9-97e7-6597244d2f78

📥 Commits

Reviewing files that changed from the base of the PR and between 1a200d8 and 8590ee4.

📒 Files selected for processing (62)

• dascore/clients/dirspool.py
• dascore/clients/filespool.py
• dascore/config.py
• dascore/core/coords.py
• dascore/core/patch.py
• dascore/core/spool.py
• dascore/core/summary.py
• dascore/exceptions.py
• dascore/io/core.py
• dascore/io/index/__init__.py
• dascore/io/index/backend.py
• dascore/io/index/catalog.py
• dascore/io/index/dialect.py
• dascore/io/index/indexer.py
• dascore/io/index/ingest.py
• dascore/io/index/lite.py
• dascore/io/index/query.py
• dascore/io/index/schema.py
• dascore/io/indexer.py
• dascore/utils/chunk.py
• dascore/utils/chunk_plan.py
• dascore/utils/coordmanager.py
• dascore/utils/hdf5.py
• dascore/utils/patch.py
• dascore/utils/paths.py
• dascore/utils/pd.py
• docs/changelog.qmd
• docs/contributing/new_format.qmd
• docs/notes/notes.qmd
• docs/notes/spool_chunking.qmd
• docs/notes/spool_index.qmd
• docs/notes/spool_selection.qmd
• docs/tutorial/file_io.qmd
• environment.yml
• pyproject.toml
• scripts/_templates/_quarto.yml
• tests/conftest.py
• tests/test_clients/test_dirspool.py
• tests/test_core/test_patch_chunk.py
• tests/test_core/test_spool.py
• tests/test_core/test_spool_select_spec.py
• tests/test_io/test_dasdae/test_dasdae.py
• tests/test_io/test_febus/test_febusg1.py
• tests/test_io/test_h5simple/test_h5simple.py
• tests/test_io/test_index/test_catalog.py
• tests/test_io/test_index/test_db_dirspool.py
• tests/test_io/test_index/test_heterogeneity_stress.py
• tests/test_io/test_index/test_index_contract.py
• tests/test_io/test_index/test_index_edge_cases.py
• tests/test_io/test_index/test_plan.py
• tests/test_io/test_index/test_schema.py
• tests/test_io/test_index/test_union.py
• tests/test_io/test_indexer.py
• tests/test_io/test_io_core.py
• tests/test_io/test_pickle/test_pickle.py
• tests/test_io/test_prodml/test_prod_ml.py
• tests/test_io/test_terra15/test_terra15.py
• tests/test_utils/test_chunk.py
• tests/test_utils/test_config.py
• tests/test_utils/test_hdf_utils.py
• tests/test_utils/test_io_utils.py
• tests/test_utils/test_patch_utils.py

💤 Files with no reviewable changes (3)

• tests/test_io/test_dasdae/test_dasdae.py
• environment.yml
• dascore/clients/filespool.py

[github-actions]

✅ Documentation built:
👉 Download
Note: You must be logged in to github and a DASDAE member to access the link.

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@dascore/utils/paths.py`:
- Around line 37-45: Wrap the cleanup call in the writable-directory probe’s
existing OSError handling so transient failures from os.remove(probe) return
False rather than escaping. Update the probe flow around probe.parent.mkdir,
open, and os.remove while preserving the existing True result only when creation
and cleanup both succeed.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 3a6e4c6e-0fb8-42dd-b3db-9153ff4bd08f

📥 Commits

Reviewing files that changed from the base of the PR and between 2cd1db4 and c3e0b87.

📒 Files selected for processing (15)

• dascore/core/spool.py
• dascore/io/index/indexer.py
• dascore/io/index/ingest.py
• dascore/utils/patch.py
• dascore/utils/paths.py
• dascore/utils/pd.py
• tests/conftest.py
• tests/test_clients/test_dirspool.py
• tests/test_core/test_spool.py
• tests/test_io/test_index/test_heterogeneity_stress.py
• tests/test_io/test_index/test_index_contract.py
• tests/test_io/test_index/test_union.py
• tests/test_utils/test_patch_utils.py
• tests/test_utils/test_paths.py
• tests/test_utils/test_pd.py

🚧 Files skipped from review as they are similar to previous changes (11)

• tests/test_utils/test_patch_utils.py
• tests/test_io/test_index/test_index_contract.py
• dascore/utils/pd.py
• dascore/io/index/indexer.py
• tests/test_core/test_spool.py
• tests/test_io/test_index/test_union.py
• tests/test_io/test_index/test_heterogeneity_stress.py
• dascore/utils/patch.py
• tests/test_clients/test_dirspool.py
• dascore/io/index/ingest.py
• dascore/core/spool.py

[coderabbitai]

Actionable comments posted: 1

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

dascore/utils/paths.py (1)

36-42: 🔒 Security & Privacy | 🟠 Major | ⚡ Quick win

Use a unique, exclusive temporary probe file.

The predictable sentinel can clobber an existing file, and open(..., "w") follows an attacker-created symlink in a writable directory. Use tempfile.mkstemp(dir=probe.parent, prefix=...) (with O_EXCL semantics) and retain the existing suppressed cleanup.

🛡️ Proposed fix

+import tempfile
+
 def directory_writable(path) -> bool:
     ...
-    name = "._dascore_write_test_delete_me"
-    probe = Path(path) / name
+    probe = None
     try:
         probe.parent.mkdir(exist_ok=True, parents=True)
-        open(probe, "w").close()
+        fd, probe_name = tempfile.mkstemp(
+            dir=probe.parent,
+            prefix="._dascore_write_test_",
+        )
+        os.close(fd)
+        probe = Path(probe_name)
     except OSError:
         return False

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@dascore/utils/paths.py` around lines 36 - 42, Replace the predictable probe
path and open call in the path-writability check with
tempfile.mkstemp(dir=probe.parent, prefix=...), retaining the existing
suppressed cleanup and closing the returned file descriptor. Preserve the
surrounding mkdir and exception handling while ensuring the probe is uniquely
created with exclusive semantics.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@tests/test_core/test_patch_chunk.py`:
- Around line 318-323: Update the _distance_envelopes helper’s zip call to pass
an explicit strict argument, using strict=True to preserve the expected
equal-length behavior while satisfying Ruff B905.

---

Outside diff comments:
In `@dascore/utils/paths.py`:
- Around line 36-42: Replace the predictable probe path and open call in the
path-writability check with tempfile.mkstemp(dir=probe.parent, prefix=...),
retaining the existing suppressed cleanup and closing the returned file
descriptor. Preserve the surrounding mkdir and exception handling while ensuring
the probe is uniquely created with exclusive semantics.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: e7c866a4-437f-4b87-ad4d-ab0d9c470bfa

📥 Commits

Reviewing files that changed from the base of the PR and between c3e0b87 and 36e639d.

📒 Files selected for processing (4)

• dascore/core/spool.py
• dascore/utils/paths.py
• tests/test_core/test_patch_chunk.py
• tests/test_core/test_spool.py

🚧 Files skipped from review as they are similar to previous changes (2)

• tests/test_core/test_spool.py
• dascore/core/spool.py