Skip to content

Commit 5522ab7

Browse files
committed
Merge Apache main into array extrema fix
2 parents 07c8de2 + 3b942e5 commit 5522ab7

669 files changed

Lines changed: 106788 additions & 8041 deletions

File tree

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.
File renamed without changes.
File renamed without changes.
File renamed without changes.
File renamed without changes.

.claude/skills/suggest-native-expression/SKILL.md renamed to .ai/skills/suggest-native-expression/SKILL.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -338,7 +338,7 @@ Steps 3 and 4 are a static assessment and need no build. That is deliberate: the
338338
cheap to run over a pool of candidates. But a `Medium` on either axis is exactly where a static
339339
read is least trustworthy, so **when either rating is Medium, gather at least one piece of
340340
empirical evidence before writing the verdict.** Building is required from here
341-
(`make`; see `CLAUDE.md`).
341+
(`make`; see `docs/source/contributor-guide/development.md`).
342342

343343
Any one of these is sufficient, cheapest first:
344344

.claude/skills/wire-datafusion-function/SKILL.md renamed to .ai/skills/wire-datafusion-function/SKILL.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -41,7 +41,7 @@ Function names must match: the string passed to `CometScalarFunction("xyz")` equ
4141

4242
### 1. Study the Spark contract
4343

44-
Find the `case class $ARGUMENTS` (PascalCase). Prefer the user's local Spark clone (`CLAUDE.md` / project memory); fall back to `git clone --depth 1 https://github.com/apache/spark.git /tmp/spark-master`. If the clone is sandbox-blocked and no local clone is found, **stop and ask** — Spark is the ground truth.
44+
Find the `case class $ARGUMENTS` (PascalCase). Prefer the user's local Spark clone (agent instructions / project memory); fall back to `git clone --depth 1 https://github.com/apache/spark.git /tmp/spark-master`. If the clone is sandbox-blocked and no local clone is found, **stop and ask** — Spark is the ground truth.
4545

4646
Note `inputTypes`, `dataType`, `eval` / `nullSafeEval`, `require` guards, ANSI mode branches, and foldable-only arguments. These define what Comet must reproduce.
4747

@@ -68,7 +68,7 @@ grep -rin "fn name" "$DF_SPARK/src/function/" 2>/dev/null | grep -i "$EXPR"
6868
grep -rin "fn name" "$DF_FUNCS/src/" 2>/dev/null | grep -i "$EXPR"
6969
```
7070

71-
**Option B — local DataFusion clone** (only if `CLAUDE.md` / memory points at one). Grep at the pinned tag — DataFusion uses lightweight tags (`<version>`, no `v` prefix), so use `tag -l` not `rev-parse --verify ^{tag}`:
71+
**Option B — local DataFusion clone** (only if agent instructions / memory point at one). Grep at the pinned tag — DataFusion uses lightweight tags (`<version>`, no `v` prefix), so use `tag -l` not `rev-parse --verify ^{tag}`:
7272

7373
```bash
7474
DF_CLONE=<path from memory>

‎.asf.yaml‎

Lines changed: 68 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -46,6 +46,19 @@ github:
4646
main:
4747
required_pull_request_reviews:
4848
required_approving_review_count: 1
49+
required_status_checks:
50+
# `Required Checks` is the aggregator job at the bottom of
51+
# .github/workflows/ci.yml. It is deliberately the only entry: the
52+
# individual heavy jobs publish a different check name depending on
53+
# whether they ran or were skipped, so none of them can be required
54+
# directly. dev/ci/check-ci-config.py keeps this string in sync with
55+
# the job's `name:`.
56+
#
57+
# Careful when editing this list. A context that never reports blocks
58+
# every merge to main, including the merge that would fix this file,
59+
# and recovering from that needs an INFRA Jira ticket.
60+
contexts:
61+
- "Required Checks"
4962
branch-0.1:
5063
required_pull_request_reviews:
5164
required_approving_review_count: 1
@@ -97,8 +110,63 @@ github:
97110
branch-0.17:
98111
required_pull_request_reviews:
99112
required_approving_review_count: 1
113+
# GitHub's merge queue. Classic `protected_branches` above has no way to
114+
# express this, but .asf.yaml rulesets accept a raw Rulesets API payload, so
115+
# no INFRA ticket is needed. Same approach as apache/iceberg-python and
116+
# apache/seatunnel; GitHub evaluates rulesets and classic branch protection
117+
# together and applies the most restrictive result.
118+
#
119+
# Only the default branch is queued. Merge queue rules do not support
120+
# wildcard ref patterns, so the branch-N.M release branches keep plain
121+
# branch protection.
122+
rulesets:
123+
- name: Merge Queue
124+
target: branch
125+
enforcement: active
126+
bypass_actors:
127+
# apache/root - ASF Infra, so a wedged queue can always be recovered.
128+
- actor_id: 118420
129+
actor_type: Team
130+
bypass_mode: always
131+
conditions:
132+
ref_name:
133+
include:
134+
- "~DEFAULT_BRANCH"
135+
exclude: []
136+
rules:
137+
- type: merge_queue
138+
parameters:
139+
# Matches enabled_merge_buttons above: squash is the only method
140+
# the repository allows.
141+
merge_method: SQUASH
142+
# ALLGREEN: every entry's own build has to be green, not just the
143+
# last one's. HEADGREEN would let an entry whose build failed ride
144+
# in on a green head, which papers over a flake but equally papers
145+
# over a real failure that a later entry happens to mask.
146+
grouping_strategy: ALLGREEN
147+
# Speculation depth: how many merge_group builds run at once. Each
148+
# queue entry gets its own build of itself plus everything ahead of
149+
# it, and merge limits do not combine those builds, so this is the
150+
# throughput dial. At 1 every merge serializes behind a ~2.5h
151+
# pipeline; 2 sustains roughly 19 merges a day, against a median of
152+
# 6 and a busiest-day 22 over the last month. Raise it if the queue
153+
# backs up, at the price of that many concurrent pipelines.
154+
max_entries_to_build: 2
155+
# These only cap how many already-green entries land in one merge;
156+
# they do not batch the builds. No minimum and no wait, so a lone
157+
# PR merges the moment its build is green.
158+
min_entries_to_merge: 1
159+
min_entries_to_merge_wait_minutes: 0
160+
max_entries_to_merge: 5
161+
# Full pipeline runs have been observed at 2h04m-2h25m, on top of
162+
# ASF runner scheduling delays. Anything under ~4h would evict
163+
# healthy entries.
164+
check_response_timeout_minutes: 300
100165
pull_requests:
101166
allow_update_branch: true
167+
# Lets a committer queue a PR before its checks have finished, which is
168+
# most of the point of having a queue.
169+
allow_auto_merge: true
102170
# publishes the content of the `asf-site` branch to
103171
# https://datafusion.apache.org/comet/
104172
publish:

‎.claude/skills‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
../.ai/skills
Lines changed: 107 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,107 @@
1+
# Licensed to the Apache Software Foundation (ASF) under one
2+
# or more contributor license agreements. See the NOTICE file
3+
# distributed with this work for additional information
4+
# regarding copyright ownership. The ASF licenses this file
5+
# to you under the Apache License, Version 2.0 (the
6+
# "License"); you may not use this file except in compliance
7+
# with the License. You may obtain a copy of the License at
8+
#
9+
# http://www.apache.org/licenses/LICENSE-2.0
10+
#
11+
# Unless required by applicable law or agreed to in writing,
12+
# software distributed under the License is distributed on an
13+
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14+
# KIND, either express or implied. See the License for the
15+
# specific language governing permissions and limitations
16+
# under the License.
17+
18+
name: "Download Artifact (with retry)"
19+
description: >
20+
Drop-in replacement for actions/download-artifact that retries the download
21+
three times. The artifact client only retries 429/500/502/503/504, so a
22+
ListArtifacts answered "(403) Forbidden: Error from intermediary" fails the
23+
step before a single byte is fetched. There is no input to widen that list,
24+
Actions has no built-in step retry, and third-party retry wrappers are not on
25+
the Apache allowed-actions list. See .github/workflows/README.md.
26+
27+
Unlike the upload wrapper, a retry here has no side effect to undo: a failed
28+
attempt leaves at most a partial extraction under `path`, which the next
29+
attempt overwrites file by file.
30+
31+
# Inputs are the subset of actions/download-artifact@v8 that this repository
32+
# uses. The boolean default has to be a real 'false' string because
33+
# core.getBooleanInput() throws on an empty value.
34+
inputs:
35+
name:
36+
description: 'Name of the artifact to download (empty means every artifact in the run)'
37+
required: false
38+
default: ''
39+
path:
40+
description: 'Destination path (empty means the workspace)'
41+
required: false
42+
default: ''
43+
pattern:
44+
description: 'Glob matching the artifacts to download; ignored when name is set'
45+
required: false
46+
default: ''
47+
merge-multiple:
48+
description: 'Extract every matched artifact straight into path rather than into one directory each'
49+
required: false
50+
default: 'false'
51+
52+
# `||` yields the first non-empty operand, so this picks whichever attempt ran.
53+
outputs:
54+
download-path:
55+
description: 'Path of artifact download'
56+
value: ${{ steps.attempt-1.outputs.download-path || steps.attempt-2.outputs.download-path || steps.attempt-3.outputs.download-path }}
57+
58+
runs:
59+
using: "composite"
60+
steps:
61+
# continue-on-error keeps a failed attempt from failing the job while still
62+
# recording outcome == 'failure', which the later attempts gate on. The last
63+
# attempt omits it so a genuinely missing artifact still fails loudly.
64+
- name: Download ${{ inputs.name || inputs.pattern || 'all artifacts' }} (attempt 1 of 3)
65+
id: attempt-1
66+
uses: actions/download-artifact@v8
67+
continue-on-error: true
68+
with:
69+
name: ${{ inputs.name }}
70+
path: ${{ inputs.path }}
71+
pattern: ${{ inputs.pattern }}
72+
merge-multiple: ${{ inputs.merge-multiple }}
73+
74+
- name: Wait before retrying ${{ inputs.name || inputs.pattern || 'all artifacts' }}
75+
if: ${{ steps.attempt-1.outcome == 'failure' }}
76+
shell: bash
77+
run: |
78+
echo "::warning::Download of '${{ inputs.name || inputs.pattern || 'all artifacts' }}' failed; retrying in 15s (attempt 2 of 3)."
79+
sleep 15
80+
81+
- name: Download ${{ inputs.name || inputs.pattern || 'all artifacts' }} (attempt 2 of 3)
82+
id: attempt-2
83+
if: ${{ steps.attempt-1.outcome == 'failure' }}
84+
uses: actions/download-artifact@v8
85+
continue-on-error: true
86+
with:
87+
name: ${{ inputs.name }}
88+
path: ${{ inputs.path }}
89+
pattern: ${{ inputs.pattern }}
90+
merge-multiple: ${{ inputs.merge-multiple }}
91+
92+
- name: Wait before final retry of ${{ inputs.name || inputs.pattern || 'all artifacts' }}
93+
if: ${{ steps.attempt-1.outcome == 'failure' && steps.attempt-2.outcome == 'failure' }}
94+
shell: bash
95+
run: |
96+
echo "::warning::Download of '${{ inputs.name || inputs.pattern || 'all artifacts' }}' failed again; retrying in 45s (attempt 3 of 3)."
97+
sleep 45
98+
99+
- name: Download ${{ inputs.name || inputs.pattern || 'all artifacts' }} (attempt 3 of 3)
100+
id: attempt-3
101+
if: ${{ steps.attempt-1.outcome == 'failure' && steps.attempt-2.outcome == 'failure' }}
102+
uses: actions/download-artifact@v8
103+
with:
104+
name: ${{ inputs.name }}
105+
path: ${{ inputs.path }}
106+
pattern: ${{ inputs.pattern }}
107+
merge-multiple: ${{ inputs.merge-multiple }}

0 commit comments

Comments
 (0)