Fixing CI failures

Author: dutow

.github/workflows/dev-setup.yml

@@ -11,6 +11,8 @@ jobs:
     runs-on: ubuntu-latest
     timeout-minutes: 20
     env:
+      # The runner has podman installed, but only the Docker daemon is running.
+      # Force the Taskfile's compose engine to docker so it doesn't pick podman.
       ENGINE: docker
 
     steps:
@@ -19,6 +21,12 @@ jobs:
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
 
+      - name: Install Task
+        uses: arduino/setup-task@v2
+        with:
+          version: 3.x
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+
       - name: Prepare dev env file
         run: |
           cp .env.development.example .env.development
@@ -29,7 +37,7 @@ jobs:
           done
 
       - name: Start dev stack
-        run: make dev-detach
+        run: task dev-detach
 
       - name: Wait for Rails to be healthy
         run: |
@@ -47,7 +55,7 @@ jobs:
           exit 1
 
       - name: Run db-reset while stack is running
-        run: make db-reset
+        run: task db-reset
 
       - name: Wait for web to restart
         run: |
@@ -65,8 +73,8 @@ jobs:
           exit 1
 
       - name: Run RSpec tests
-        run: make test
+        run: task test
 
       - name: Teardown
         if: always()
-        run: make down
+        run: task down

Taskfile.yml

@@ -2,7 +2,7 @@ version: '3'
 
 vars:
   ENGINE:
-    sh: command -v podman >/dev/null 2>&1 && echo podman || echo docker
+    sh: echo "${ENGINE:-$(command -v podman >/dev/null 2>&1 && echo podman || echo docker)}"
   COMPOSE: '{{.ENGINE}} compose --env-file .env.development -f compose.dev.yml'
   PSQL_IMPORT: "{{.COMPOSE}} exec -T db bash -lc 'psql -v ON_ERROR_STOP=1 -U ${POSTGRES_USER:-hackorum} -d ${POSTGRES_DB:-hackorum_development}'"
 
@@ -54,10 +54,14 @@ tasks:
       - '{{.COMPOSE}} exec web bin/rails console'
 
   test:
-    desc: Run RSpec in the web container (uses test database). Pass args after --, e.g. task test -- spec/models
-    cmds:
-      - '{{.COMPOSE}} exec -e RAILS_ENV=test -e DATABASE_URL=postgresql://hackorum:hackorum@db:5432/hackorum_test web bin/rails db:prepare'
-      - '{{.COMPOSE}} exec -e RAILS_ENV=test -e DATABASE_URL=postgresql://hackorum:hackorum@db:5432/hackorum_test web bundle exec rspec {{.CLI_ARGS}}'
+    desc: Run RSpec in a one-off container (uses test database). Pass args after --, e.g. task test -- spec/models
+    cmds:
+      # Use `run --rm` (a dedicated throwaway container) rather than `exec` into the
+      # running dev server: that avoids loading rspec on top of the live bin/dev
+      # process tree, where a foreman/server crash or an OOM victim selection would
+      # take the exec'd rspec down with it (SIGKILL/exit 137 in CI).
+      - '{{.COMPOSE}} run --rm -e RAILS_ENV=test -e DATABASE_URL=postgresql://hackorum:hackorum@db:5432/hackorum_test web bin/rails db:prepare'
+      - '{{.COMPOSE}} run --rm -e RAILS_ENV=test -e DATABASE_URL=postgresql://hackorum:hackorum@db:5432/hackorum_test web bundle exec rspec {{.CLI_ARGS}}'
 
   db-migrate:
     desc: Run db:migrate

app/services/topic_candidate_search.rb

@@ -51,7 +51,7 @@ def scope
 
   def rank_select
     return "0 AS relevance_score" if @q.blank?
-    Topic.sanitize_sql_array(["ts_rank(topics.title_tsv, websearch_to_tsquery('english', ?)) AS relevance_score", @q])
+    Topic.sanitize_sql_array([ "ts_rank(topics.title_tsv, websearch_to_tsquery('english', ?)) AS relevance_score", @q ])
   end
 
   def apply_text(relation)

script/hackorum_commits.rb

@@ -105,9 +105,9 @@ def initialize(path)
     def upsert_commit(sha:, subject:, body:, authored_at:, committed_at:,
                       author_name:, author_email:, committer_name:, committer_email:,
                       branches:, versions:, stage: "walked")
-      params = [sha, subject, body, authored_at, committed_at,
+      params = [ sha, subject, body, authored_at, committed_at,
                 author_name, author_email, committer_name, committer_email,
-                JSON.generate(branches), JSON.generate(versions), stage, Time.now.utc.iso8601]
+                JSON.generate(branches), JSON.generate(versions), stage, Time.now.utc.iso8601 ]
       @db.execute(<<~SQL, params)
         INSERT INTO commits (sha, subject, body, authored_at, committed_at,
           author_name, author_email, committer_name, committer_email,
@@ -119,96 +119,96 @@ def upsert_commit(sha:, subject:, body:, authored_at:, committed_at:,
     end
 
     def commit(sha)
-      @db.execute("SELECT * FROM commits WHERE sha = ?", [sha]).first
+      @db.execute("SELECT * FROM commits WHERE sha = ?", [ sha ]).first
     end
 
     def set_stage(sha, stage)
       @db.execute("UPDATE commits SET stage = ?, updated_at = ? WHERE sha = ?",
-                  [stage, Time.now.utc.iso8601, sha])
+                  [ stage, Time.now.utc.iso8601, sha ])
     end
 
     def add_relation(from_sha:, to_sha:, kind:, confidence:, method:)
-      @db.execute(<<~SQL, [from_sha, to_sha, kind, confidence, method])
+      @db.execute(<<~SQL, [ from_sha, to_sha, kind, confidence, method ])
         INSERT OR IGNORE INTO commit_relations (from_sha, to_sha, kind, confidence, method)
         VALUES (?,?,?,?,?)
       SQL
     end
 
     def relations_for(sha)
-      @db.execute("SELECT * FROM commit_relations WHERE from_sha = ?", [sha])
+      @db.execute("SELECT * FROM commit_relations WHERE from_sha = ?", [ sha ])
     end
 
     def api_cache_get(key)
-      row = @db.execute("SELECT response FROM api_cache WHERE request_key = ?", [key]).first
+      row = @db.execute("SELECT response FROM api_cache WHERE request_key = ?", [ key ]).first
       row && row["response"]
     end
 
     def api_cache_put(key, response)
       @db.execute("INSERT OR REPLACE INTO api_cache (request_key, response, fetched_at) VALUES (?,?,?)",
-                  [key, response, Time.now.utc.iso8601])
+                  [ key, response, Time.now.utc.iso8601 ])
     end
 
     def add_candidate(sha:, topic_id:, source:, prefilter_score:, metadata:)
-      @db.execute(<<~SQL, [sha, topic_id, source, prefilter_score, JSON.generate(metadata)])
+      @db.execute(<<~SQL, [ sha, topic_id, source, prefilter_score, JSON.generate(metadata) ])
         INSERT OR REPLACE INTO thread_candidates (sha, topic_id, source, prefilter_score, metadata)
         VALUES (?,?,?,?,?)
       SQL
     end
 
     def candidates_for(sha)
-      @db.execute("SELECT * FROM thread_candidates WHERE sha = ?", [sha])
+      @db.execute("SELECT * FROM thread_candidates WHERE sha = ?", [ sha ])
     end
 
     def llm_cache_get(key)
-      row = @db.execute("SELECT response FROM llm_cache WHERE prompt_hash = ?", [key]).first
+      row = @db.execute("SELECT response FROM llm_cache WHERE prompt_hash = ?", [ key ]).first
       row && row["response"]
     end
 
     def llm_cache_put(key, response, model)
       @db.execute("INSERT OR REPLACE INTO llm_cache (prompt_hash, response, model, created_at) VALUES (?,?,?,?)",
-                  [key, response, model, Time.now.utc.iso8601])
+                  [ key, response, model, Time.now.utc.iso8601 ])
     end
 
     def add_link(sha:, topic_id:, mailing_list:, method:, confidence:, verdict:, evidence:, external_message_id:)
-      @db.execute(<<~SQL, [sha, topic_id, mailing_list, method, confidence, verdict, evidence, external_message_id])
+      @db.execute(<<~SQL, [ sha, topic_id, mailing_list, method, confidence, verdict, evidence, external_message_id ])
         INSERT OR REPLACE INTO thread_links
           (sha, topic_id, mailing_list, method, confidence, verdict, evidence, external_message_id)
         VALUES (?,?,?,?,?,?,?,?)
       SQL
     end
 
     def links_for(sha)
-      @db.execute("SELECT * FROM thread_links WHERE sha = ?", [sha])
+      @db.execute("SELECT * FROM thread_links WHERE sha = ?", [ sha ])
     end
 
     def add_fact(sha:, kind:, value:, method:, confidence:, evidence:)
-      @db.execute(<<~SQL, [sha, kind, value, method, confidence, evidence])
+      @db.execute(<<~SQL, [ sha, kind, value, method, confidence, evidence ])
         INSERT OR IGNORE INTO commit_facts (sha, kind, value, method, confidence, evidence)
         VALUES (?,?,?,?,?,?)
       SQL
     end
 
     def facts_for(sha)
-      @db.execute("SELECT * FROM commit_facts WHERE sha = ?", [sha])
+      @db.execute("SELECT * FROM commit_facts WHERE sha = ?", [ sha ])
     end
 
     def each_commit
       @db.execute("SELECT * FROM commits ORDER BY committed_at").each { |row| yield row }
     end
 
     def each_commit_at_stage(stage)
-      @db.execute("SELECT * FROM commits WHERE stage = ? ORDER BY committed_at", [stage]).each { |r| yield r }
+      @db.execute("SELECT * FROM commits WHERE stage = ? ORDER BY committed_at", [ stage ]).each { |r| yield r }
     end
 
     def set_message_ids(sha, ids)
       ids.each do |id|
         @db.execute("INSERT OR IGNORE INTO commit_facts (sha, kind, value, method, confidence, evidence) VALUES (?,?,?,?,?,?)",
-                    [sha, "discussion_message_id", id, "trailer", 1.0, "Discussion: #{id}"])
+                    [ sha, "discussion_message_id", id, "trailer", 1.0, "Discussion: #{id}" ])
       end
     end
 
     def message_ids_for(sha)
-      @db.execute("SELECT value FROM commit_facts WHERE sha = ? AND kind = 'discussion_message_id'", [sha])
+      @db.execute("SELECT value FROM commit_facts WHERE sha = ? AND kind = 'discussion_message_id'", [ sha ])
          .map { |r| r["value"] }
     end
   end
@@ -314,7 +314,7 @@ def default_refs
       out = git("for-each-ref", "--format=%(refname:short)", "refs/remotes/origin")
       rel = out.lines.map(&:strip).map { |r| r.sub(%r{\Aorigin/}, "") }
                .select { |r| r == "master" || r.start_with?("REL") }
-      (["master"] + rel).uniq
+      ([ "master" ] + rel).uniq
     end
 
     def version_for_ref(ref)
@@ -357,7 +357,7 @@ def detect_backport_twins(shas)
         row = @store.commit(sha)
         next unless row
         next if row["subject"].to_s.strip.length < 10 # too generic to match safely
-        key = [row["subject"], row["author_email"]]
+        key = [ row["subject"], row["author_email"] ]
         by_key[key] << row
       end
 
@@ -376,10 +376,10 @@ def detect_backport_twins(shas)
 
     def list_shas(ref)
       args = if ref_exists?("origin/#{ref}")
-               ["rev-list", "origin/#{ref}"]
-             else
-               ["rev-list", ref]
-             end
+               [ "rev-list", "origin/#{ref}" ]
+      else
+               [ "rev-list", ref ]
+      end
       args << "--max-count=#{@limit}" if @limit
       args << "--since=#{@since}" if @since
       git(*args).lines.map(&:strip).reject(&:empty?)
@@ -509,7 +509,7 @@ def discover(commit, facts:, message_ids:)
       from, to = date_window(commit["committed_at"])
 
       search_added = 0
-      [true, false].each do |patches_only|
+      [ true, false ].each do |patches_only|
         hits = @api.search_candidates(q: terms.join(" "), from: from, to: to,
                                       mailing_lists: [], patches_only: patches_only,
                                       limit: @config.candidate_limit)
@@ -626,7 +626,7 @@ def parse_meta(candidate)
     end
 
     def build_user_prompt(commit, candidates)
-      lines = ["COMMIT subject: #{commit['subject']}", "COMMIT body:", commit["body"].to_s, "", "CANDIDATE THREADS:"]
+      lines = [ "COMMIT subject: #{commit['subject']}", "COMMIT body:", commit["body"].to_s, "", "CANDIDATE THREADS:" ]
       candidates.each do |c|
         meta = parse_meta(c)
         lines << "- topic_id=#{c['topic_id']} lists=#{Array(meta['mailing_lists']).join(',')} " \

spec/scripts/hackorum_commits/api_client_spec.rb

@@ -8,7 +8,7 @@
   it "fetches and parses candidate search, caching the second call" do
     stub = stub_request(:get, "http://hk.test/topics/search_candidates.json")
            .with(query: hash_including("q" => "vacuum"))
-           .to_return(status: 200, body: { candidates: [{ topic_id: 7, title: "T" }] }.to_json,
+           .to_return(status: 200, body: { candidates: [ { topic_id: 7, title: "T" } ] }.to_json,
                       headers: { "Content-Type" => "application/json" })
 
     first = client.search_candidates(q: "vacuum", from: "2020-01-01T00:00:00Z",
@@ -25,7 +25,7 @@
 
   it "resolves a message-id, returning nil on 404" do
     stub_request(:get, %r{http://hk\.test/messages/by-id/.*\.json})
-      .to_return(status: 200, body: { topic_id: 9, mailing_lists: ["pgsql-hackers"] }.to_json,
+      .to_return(status: 200, body: { topic_id: 9, mailing_lists: [ "pgsql-hackers" ] }.to_json,
                  headers: { "Content-Type" => "application/json" })
     expect(client.resolve_message_id("abc@x")["topic_id"]).to eq(9)
 

spec/scripts/hackorum_commits/confirmer_spec.rb

@@ -8,8 +8,8 @@
   let(:commit) { { "sha" => "abc", "subject" => "Fix vacuum", "body" => "details" } }
   let(:candidates) do
     [
-      { "topic_id" => 100, "source" => "trailer", "metadata" => { "mailing_lists" => ["pgsql-hackers"] }.to_json },
-      { "topic_id" => 200, "source" => "search", "metadata" => { "title" => "vacuum perf", "mailing_lists" => ["pgsql-bugs"] }.to_json }
+      { "topic_id" => 100, "source" => "trailer", "metadata" => { "mailing_lists" => [ "pgsql-hackers" ] }.to_json },
+      { "topic_id" => 200, "source" => "search", "metadata" => { "title" => "vacuum perf", "mailing_lists" => [ "pgsql-bugs" ] }.to_json }
     ]
   end
 
@@ -18,7 +18,7 @@
       "links" => [
         { "topic_id" => 200, "verdict" => "related", "confidence" => 0.8, "evidence" => "same crash" }
       ],
-      "facts" => [{ "kind" => "cve", "value" => "CVE-2025-9999" }]
+      "facts" => [ { "kind" => "cve", "value" => "CVE-2025-9999" } ]
     )
 
     confirmer.confirm(commit, candidates: candidates)

spec/scripts/hackorum_commits/discovery_spec.rb

@@ -7,7 +7,7 @@
   subject(:discovery) { described_class.new(config: config, store: store, api: api) }
 
   it "strips boilerplate and includes contributor names in search terms" do
-    facts = [{ kind: "author", value: "Dan Dev <dan@x>" }, { kind: "reported_by", value: "Bob <bob@x>" }]
+    facts = [ { kind: "author", value: "Dan Dev <dan@x>" }, { kind: "reported_by", value: "Bob <bob@x>" } ]
     terms = discovery.search_terms(subject: "Fix vacuum throttling bug", facts: facts)
     expect(terms).to include("vacuum", "throttling")
     expect(terms).not_to include("Fix")
@@ -23,13 +23,13 @@
 
   it "records trailer and search candidates, keeping trailers past the cap" do
     allow(api).to receive(:resolve_message_id).with("disc@x")
-      .and_return({ "topic_id" => 100, "mailing_lists" => ["pgsql-hackers"] })
+      .and_return({ "topic_id" => 100, "mailing_lists" => [ "pgsql-hackers" ] })
     allow(api).to receive(:search_candidates).and_return(
       (1..10).map { |i| { "topic_id" => i, "title" => "t#{i}" } }
     )
 
     commit = { "sha" => "abc", "subject" => "Fix vacuum", "committed_at" => "2020-06-01T00:00:00Z" }
-    discovery.discover(commit, facts: [], message_ids: ["disc@x"])
+    discovery.discover(commit, facts: [], message_ids: [ "disc@x" ])
 
     rows = store.candidates_for("abc")
     trailer = rows.select { |r| r["source"] == "trailer" }
@@ -40,7 +40,7 @@
 
   it "trailer candidate wins over search when the same topic_id appears in both" do
     allow(api).to receive(:resolve_message_id).with("shared@x")
-      .and_return({ "topic_id" => 5, "mailing_lists" => ["pgsql-hackers"] })
+      .and_return({ "topic_id" => 5, "mailing_lists" => [ "pgsql-hackers" ] })
     allow(api).to receive(:search_candidates).and_return(
       [
         { "topic_id" => 5,  "score" => 0.3, "title" => "overlap" },
@@ -50,7 +50,7 @@
     )
 
     commit = { "sha" => "def", "subject" => "Fix index vacuum bug", "committed_at" => "2020-06-01T00:00:00Z" }
-    discovery.discover(commit, facts: [], message_ids: ["shared@x"])
+    discovery.discover(commit, facts: [], message_ids: [ "shared@x" ])
 
     rows = store.candidates_for("def")
     topic5 = rows.find { |r| r["topic_id"] == 5 }
@@ -65,12 +65,12 @@
 
   it "trailer pre-seed does not reduce the search candidate budget" do
     allow(api).to receive(:resolve_message_id).with("trailer@x")
-      .and_return({ "topic_id" => 99, "mailing_lists" => ["pgsql-hackers"] })
+      .and_return({ "topic_id" => 99, "mailing_lists" => [ "pgsql-hackers" ] })
     search_hits = (1..10).map { |i| { "topic_id" => i, "score" => 0.5, "title" => "hit#{i}" } }
     allow(api).to receive(:search_candidates).and_return(search_hits, [])
 
     commit = { "sha" => "ghi", "subject" => "Fix index vacuum scan", "committed_at" => "2020-06-01T00:00:00Z" }
-    discovery.discover(commit, facts: [], message_ids: ["trailer@x"])
+    discovery.discover(commit, facts: [], message_ids: [ "trailer@x" ])
 
     rows = store.candidates_for("ghi")
     trailer = rows.select { |r| r["source"] == "trailer" }

spec/scripts/hackorum_commits/exporter_spec.rb

@@ -9,7 +9,7 @@
                         authored_at: "2020-01-01", committed_at: "2020-01-02",
                         author_name: "Dan", author_email: "dan@x",
                         committer_name: "Andrew", committer_email: "a@x",
-                        branches: ["master", "REL_17_STABLE"], versions: ["devel", "17"])
+                        branches: [ "master", "REL_17_STABLE" ], versions: [ "devel", "17" ])
     store.add_fact(sha: "abc", kind: "reviewer", value: "Jane", method: "trailer", confidence: 1.0, evidence: "Reviewed-by: Jane")
     store.add_link(sha: "abc", topic_id: 100, mailing_list: "pgsql-hackers", method: "trailer",
                    confidence: 1.0, verdict: "related", evidence: "trailer", external_message_id: "m@x")
@@ -27,7 +27,7 @@
     expect(rec["versions"]).to include("17")
     expect(rec["committer"]["name"]).to eq("Andrew")
     expect(rec["facts"].map { |f| f["kind"] }).to include("reviewer")
-    expect(rec["thread_links"].map { |l| l["topic_id"] }).to eq([100])
+    expect(rec["thread_links"].map { |l| l["topic_id"] }).to eq([ 100 ])
     expect(rec["generator"]["tool_version"]).to eq(HackorumCommits::VERSION)
     expect(rec["generator"]["model"]).to eq("qwen")
   end