You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
test(operators): several columns, names, mappings and inputs, in every order, role and arrangement (ITL-1236, part 1) - #442
Merge order: #442, then #443, then #444. All three target main; each contains the previous one's commits. After one is merged, main is merged into the next (never a rebase), so its diff shrinks to its own files.
Part 1 of ITL-1236 (operators). Related: ITL-1241, ITL-1242, ITL-1243, ITL-1247, ITL-1249 (defects and a question found here, tested as xfail(strict=True)), ITL-1226 (PR #434, GroupBy, not duplicated here).
Why
GroupBy was tested with one tag and one data column, which hid a result that depended on the order of the input's tag columns (ITL-1226). This PR audits every argument that takes several items and closes the gaps for the operators.
What
Inventory, docs/dev/specs/itl-1236-several-item-test-audit.md: one row per argument or input that takes several items (operators, sources, function pods, binding, ColumnConfig), with the existing test that covers each of (a) several items, (b) several orders, (c) each role, (d) 2+ tags and 2+ data in several arrangements, (e) sync and async, and the ranked gaps.
Shared inputs, tests/test_core/operators/several_items.py: MAIN (tags t1, t2, t3; data d1, d2, d3) and join partners P, Q, R. Each can be built in any tag order, in three table layouts, with plain tags, data-tags or a mix.
Tests, tests/test_core/operators/test_several_items.py, public interface only (op(stream), keys, iter_data, hashes), every case on the sync and the async path:
Every operator gives the same rows, row hashes, system tags, source info, content and pipeline hashes for every arrangement of its input (27 arrangements).
The order of a set argument changes nothing, not even the column order or the operator's identity: select and drop names (tag, data, both), DropNulls names, MapTags / MapData / MapColumns mappings (swaps included), PolarsFilter constraints. PolarsFilter predicates in every order keep the same rows.
Join over 2, 3 and 4 inputs: every permutation gives the same rows, hashes, system-tag names and column order. MergeJoin both ways with three colliding columns. SemiJoin matches on every shared tag.
Names of several roles in one call (plain tag, data-tag, data) follow the documented role rules.
AddTags: the order of its names is the order of the new tags and part of its identity (as documented).
Rule: CLAUDE.md and .zed/rules, "Testing conventions".
Defects found (not fixed here)
Test (xfail(strict=True))
Issue
test_merge_join_of_tied_values_does_not_depend_on_the_order_of_its_inputs: tied values keep their source info in argument order
ITL-1241
test_map_tags_swaps_data_tags_as_it_swaps_plain_tags: a swap of data-tags is refused
ITL-1242
test_merge_join_of_a_nullable_column_holding_nulls: TypeError on a null in a merged column
ITL-1243
test_unary_result_keys_equal_its_row_keys_in_every_arrangement, test_*_async_column_order_equals_sync*: on an input whose table layout differs from its tag order, a result's keys disagree with its rows, and async rows' column order with sync rows' (no hash, no stored table: an operator node stores one order on both orchestrators, tested)
ITL-1249
test_polars_filter_predicate_order_is_not_identity: the order of the predicates is in the identity (rule undecided; the test states the proposal)
ITL-1247
No library code changes; no golden file changes; no changelog fragment (tests and developer docs only, per changelog.d/README.md).
Follow-up PRs on ITL-1236: sources, function pods and binding.
Posted by a Claude Code agent on behalf of @eywalker.
…y order, role and arrangement (ITL-1236)
A shared family of inputs with three tags and three data columns, in several tag
orders, table layouts and role settings (plain, data-tags, mixed), and tests of every
operator over it, sync and async: the result does not depend on the arrangement of the
input; the order of a set argument (names, rename mappings, constraints) changes
nothing; Join over 2, 3 and 4 inputs and MergeJoin are commutative; SemiJoin matches
on every shared tag. Defects found are xfail(strict=True): ITL-1241 (MergeJoin ties),
ITL-1242 (MapTags swap of data-tags), MergeJoin null crash (issue pending).
The inventory of every several-item argument, with the evidence and the ranked gaps,
is docs/dev/specs/itl-1236-several-item-test-audit.md. CLAUDE.md and .zed/rules state
the rule in "Testing conventions".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Limit details: You’ve used the included review currently available. Your 115 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.
The reason will be displayed to describe this comment to others. Learn more.
🟡 Changes recommended
The join tests leave coverage gaps: async results lack sync-equivalence assertions, restricted-input behavior is tested only synchronously, and MergeJoin omits reordered and interleaved column layouts.
This first part of ITL-1236 expands operator regression coverage for multiple columns, argument orders, column roles and input layouts, complementing the GroupBy work in PR #434. It changes tests and developer guidance only, not library code or golden hashes.
Changes:
Adds reusable multi-column inputs and sync/async operator test matrices.
Records coverage gaps and known defects, with strict expected-failure tests.
Adds matching several-item testing guidance to both agent instruction files.
…se; MergeJoin and the restricted SemiJoin in every layout (ITL-1236)
Addresses the Copilot review of #442.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The reason will be displayed to describe this comment to others. Learn more.
🟡 Changes recommended
The new tests can miss incorrect predicate-combination semantics and sync/async column-order mismatches because they lack expected-row and order-sensitive assertions.
This only checks agreement between permutations, so an implementation that combines multiple predicates with OR instead of AND still passes: all six permutations would return every MAIN row, whereas the intended conjunction keeps only the row with d1 == 50. The other new predicate cases likewise compare arrangements or sync/async results without an expected-row assertion. Please assert the expected surviving row here so the several-predicate coverage also detects incorrect combination semantics.
…arrangement; defect 7 as xfail (ITL-1236)
Addresses the second Copilot review of #442: sync and async rows are compared with
their column order, system-tag order and Arrow schemas within each arrangement, and
a result's keys with its rows' keys. Both fail on inputs whose table layout differs
from the tag order (reported, issue pending); those combinations are
xfail(strict=True), the others are asserted.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The reason will be displayed to describe this comment to others. Learn more.
🟡 Changes recommended
The tests leave order independence during actual tag selection unverified, while broad xfail sweeps can mask regressions in previously working arrangements.
This case selects all three tags, so every permutation takes the passthrough branch in SelectTagColumns.unary_static_process, and the async path has no tags to drop. The partial selections elsewhere use only one argument order, leaving order independence during actual selection untested. Select two of the three tags here so the permutation matrix exercises selection on both paths and in every role.
The reason will be displayed to describe this comment to others. Learn more.
🔵 Needs a closer look
Human assessment is needed because mixed-role data-tag handling lacks test coverage across argument orders and input arrangements on both synchronous and asynchronous paths.
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
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.
Merge order: #442, then #443, then #444. All three target
main; each contains the previous one's commits. After one is merged,mainis merged into the next (never a rebase), so its diff shrinks to its own files.Part 1 of ITL-1236 (operators). Related: ITL-1241, ITL-1242, ITL-1243, ITL-1247, ITL-1249 (defects and a question found here, tested as
xfail(strict=True)), ITL-1226 (PR #434,GroupBy, not duplicated here).Why
GroupBywas tested with one tag and one data column, which hid a result that depended on the order of the input's tag columns (ITL-1226). This PR audits every argument that takes several items and closes the gaps for the operators.What
docs/dev/specs/itl-1236-several-item-test-audit.md: one row per argument or input that takes several items (operators, sources, function pods, binding,ColumnConfig), with the existing test that covers each of (a) several items, (b) several orders, (c) each role, (d) 2+ tags and 2+ data in several arrangements, (e) sync and async, and the ranked gaps.tests/test_core/operators/several_items.py:MAIN(tagst1,t2,t3; datad1,d2,d3) and join partnersP,Q,R. Each can be built in any tag order, in three table layouts, with plain tags, data-tags or a mix.tests/test_core/operators/test_several_items.py, public interface only (op(stream),keys,iter_data, hashes), every case on the sync and the async path:DropNullsnames,MapTags/MapData/MapColumnsmappings (swaps included),PolarsFilterconstraints.PolarsFilterpredicates in every order keep the same rows.Joinover 2, 3 and 4 inputs: every permutation gives the same rows, hashes, system-tag names and column order.MergeJoinboth ways with three colliding columns.SemiJoinmatches on every shared tag.AddTags: the order of its names is the order of the new tags and part of its identity (as documented).CLAUDE.mdand.zed/rules, "Testing conventions".Defects found (not fixed here)
xfail(strict=True))test_merge_join_of_tied_values_does_not_depend_on_the_order_of_its_inputs: tied values keep their source info in argument ordertest_map_tags_swaps_data_tags_as_it_swaps_plain_tags: a swap of data-tags is refusedtest_merge_join_of_a_nullable_column_holding_nulls:TypeErroron a null in a merged columntest_unary_result_keys_equal_its_row_keys_in_every_arrangement,test_*_async_column_order_equals_sync*: on an input whose table layout differs from its tag order, a result's keys disagree with its rows, and async rows' column order with sync rows' (no hash, no stored table: an operator node stores one order on both orchestrators, tested)test_polars_filter_predicate_order_is_not_identity: the order of the predicates is in the identity (rule undecided; the test states the proposal)No library code changes; no golden file changes; no changelog fragment (tests and developer docs only, per
changelog.d/README.md).Follow-up PRs on ITL-1236: sources, function pods and binding.
Posted by a Claude Code agent on behalf of @eywalker.
🤖 Generated with Claude Code