Merge 64d1d33 into 3aec49f

conbench/api/_examples.py

@@ -228,11 +228,13 @@ def _api_distribution_entity(
     distribution_id,
     case_id,
     context_id,
+    commit_id,
 ):
     result = {
         "id": distribution_id,
         "case_id": case_id,
         "context_id": context_id,
+        "commit_id": commit_id,
         "machine_hash": "diana-2-4-17179869184",
         "unit": "s",
         "mean_mean": "0.036369",
@@ -364,6 +366,7 @@ def _api_run_entity(run_id, commit_id, machine_id, now, baseline_id):
     "some-distribution-uuid-1",
     "some-case-uuid-1",
     "some-context-uuid-1",
+    "some-commit-uuid-1",
 )
 HISTORY_ENTITY = _api_history_entity(
     "some-benchmark-uuid-1",

conbench/entities/distribution.py

@@ -25,6 +25,7 @@ class Distribution(Base, EntityMixin):
     repository = NotNull(s.String(100))
     case_id = NotNull(s.String(50), s.ForeignKey("case.id", ondelete="CASCADE"))
     context_id = NotNull(s.String(50), s.ForeignKey("context.id", ondelete="CASCADE"))
+    commit_id = Nullable(s.String(50), s.ForeignKey("commit.id", ondelete="CASCADE"))
     machine_hash = NotNull(s.String(250))
     unit = NotNull(s.Text)
     mean_mean = Nullable(s.Numeric, check("mean_mean>=0"))
@@ -53,6 +54,7 @@ class Distribution(Base, EntityMixin):
 s.Index("distribution_repository_index", Distribution.repository)
 s.Index("distribution_case_id_index", Distribution.case_id)
 s.Index("distribution_context_id_index", Distribution.context_id)
+s.Index("distribution_commit_id_index", Distribution.commit_id)
 s.Index("distribution_machine_hash_index", Distribution.machine_hash)
 
 
@@ -67,6 +69,7 @@ def _dump(self, distribution):
             "repository": distribution.repository,
             "case_id": distribution.case_id,
             "context_id": distribution.context_id,
+            "commit_id": distribution.commit_id,
             "machine_hash": distribution.machine_hash,
             "unit": distribution.unit,
             "mean_mean": self.decimal_fmt.format(distribution.mean_mean),
@@ -90,6 +93,7 @@ def get_distribution_history(case_id, context_id, machine_hash):
             Distribution.sha,
             Distribution.case_id,
             Distribution.context_id,
+            Distribution.commit_id,
             Distribution.machine_hash,
             Distribution.unit,
             Distribution.mean_mean,
@@ -127,16 +131,19 @@ def get_commits_up(repository, sha, limit):
     return Session.query(index).filter(index.c.row_number >= n).limit(limit)
 
 
-def get_distribution(repository, sha, case_id, context_id, machine_hash, limit):
+def get_distribution(
+    repository, sha, case_id, context_id, commit_id, machine_hash, limit
+):
     from ..entities.summary import Summary
 
     commits_up = get_commits_up(repository, sha, limit).subquery().alias("commits_up")
     return (
         Session.query(
             func.text(repository).label("repository"),
             func.text(sha).label("sha"),
-            Summary.case_id,
-            Summary.context_id,
+            func.text(case_id).label("case_id"),
+            func.text(context_id).label("context_id"),
+            func.text(commit_id).label("commit_id"),
             Machine.hash,
             func.max(Summary.unit).label("unit"),
             func.avg(Summary.mean).label("mean_mean"),
@@ -179,6 +186,7 @@ def update_distribution(repository, sha, summary, limit):
         sha,
         summary.case_id,
         summary.context_id,
+        summary.run.commit_id,
         summary.run.machine.hash,
         limit,
     ).first()

conbench/tests/api/_expected_docs.py

@@ -462,6 +462,7 @@
                         "example": [
                             {
                                 "case_id": "some-case-uuid-1",
+                                "commit_id": "some-commit-uuid-1",
                                 "context_id": "some-context-uuid-1",
                                 "first_timestamp": "2021-02-25T01:02:51",
                                 "id": "some-distribution-uuid-1",

conbench/tests/api/test_distribution.py

@@ -9,6 +9,7 @@ def _expected_entity(distribution):
         distribution.id,
         distribution.case_id,
         distribution.context_id,
+        distribution.commit_id,
     )
 
 

conbench/tests/entities/test_distribution.py

@@ -57,7 +57,7 @@
 (SELECT commit.id AS id, commit.sha AS sha, commit.timestamp AS timestamp 
 FROM commit 
 WHERE commit.repository = :repository_1 ORDER BY commit.timestamp DESC)
- SELECT text(:text_1) AS repository, text(:text_2) AS sha, summary.case_id, summary.context_id, concat(machine.name, :concat_1, machine.cpu_core_count, :concat_2, machine.cpu_thread_count, :concat_3, machine.memory_bytes) AS hash, max(summary.unit) AS unit, avg(summary.mean) AS mean_mean, stddev(summary.mean) AS mean_sd, avg(summary.min) AS min_mean, stddev(summary.min) AS min_sd, avg(summary.max) AS max_mean, stddev(summary.max) AS max_sd, avg(summary.median) AS median_mean, stddev(summary.median) AS median_sd, min(commits_up.timestamp) AS first_timestamp, max(commits_up.timestamp) AS last_timestamp, count(summary.mean) AS observations 
+ SELECT text(:text_1) AS repository, text(:text_2) AS sha, text(:text_3) AS case_id, text(:text_4) AS context_id, text(:text_5) AS commit_id, concat(machine.name, :concat_1, machine.cpu_core_count, :concat_2, machine.cpu_thread_count, :concat_3, machine.memory_bytes) AS hash, max(summary.unit) AS unit, avg(summary.mean) AS mean_mean, stddev(summary.mean) AS mean_sd, avg(summary.min) AS min_mean, stddev(summary.min) AS min_sd, avg(summary.max) AS max_mean, stddev(summary.max) AS max_sd, avg(summary.median) AS median_mean, stddev(summary.median) AS median_sd, min(commits_up.timestamp) AS first_timestamp, max(commits_up.timestamp) AS last_timestamp, count(summary.mean) AS observations 
 FROM summary JOIN run ON run.id = summary.run_id JOIN machine ON machine.id = run.machine_id JOIN (SELECT commit_index.id AS id, commit_index.sha AS sha, commit_index.timestamp AS timestamp, commit_index.row_number AS row_number 
 FROM (SELECT ordered_commits.id AS id, ordered_commits.sha AS sha, ordered_commits.timestamp AS timestamp, row_number() OVER () AS row_number 
 FROM ordered_commits) AS commit_index 
@@ -132,7 +132,9 @@ def test_distribution_queries():
     assert query == ROW_NUMBER
     query = str(get_commits_up(REPO, "SHA", 3).statement.compile())
     assert query == COMMITS_UP
-    query = str(get_distribution(REPO, "SHA", "ID", "ID", "ID", 3).statement.compile())
+    query = str(
+        get_distribution(REPO, "SHA", "ID", "ID", "ID", "ID", 3).statement.compile()
+    )
     assert query == DISTRIBUTION
 
 
@@ -302,13 +304,14 @@ def test_distribution():
     # ----- get_distribution
 
     assert get_distribution(
-        REPO, "55555", case_id, context_id, machine_hash, 10
+        REPO, "55555", case_id, context_id, commit_5.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "55555",
             summary_5.case_id,
             summary_5.context_id,
+            commit_5.id,
             MACHINE,
             "s",
             decimal.Decimal("1.8440000000000000"),
@@ -325,13 +328,14 @@ def test_distribution():
         )
     ]
     assert get_distribution(
-        REPO, "44444", case_id, context_id, machine_hash, 10
+        REPO, "44444", case_id, context_id, commit_4.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "44444",
             summary_4.case_id,
             summary_4.context_id,
+            commit_4.id,
             MACHINE,
             "s",
             decimal.Decimal("1.5475000000000000"),
@@ -348,13 +352,14 @@ def test_distribution():
         )
     ]
     assert get_distribution(
-        REPO, "33333", case_id, context_id, machine_hash, 10
+        REPO, "33333", case_id, context_id, commit_3.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "33333",
             summary_3.case_id,
             summary_3.context_id,
+            commit_3.id,
             MACHINE,
             "s",
             decimal.Decimal("1.6966666666666667"),
@@ -371,13 +376,14 @@ def test_distribution():
         )
     ]
     assert get_distribution(
-        REPO, "22222", case_id, context_id, machine_hash, 10
+        REPO, "22222", case_id, context_id, commit_2.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "22222",
             summary_2.case_id,
             summary_2.context_id,
+            commit_2.id,
             MACHINE,
             "s",
             decimal.Decimal("2.0300000000000000"),
@@ -394,13 +400,14 @@ def test_distribution():
         )
     ]
     assert get_distribution(
-        REPO, "11111", case_id, context_id, machine_hash, 10
+        REPO, "11111", case_id, context_id, commit_1.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "11111",
             summary_1.case_id,
             summary_1.context_id,
+            commit_1.id,
             MACHINE,
             "s",
             decimal.Decimal("2.0300000000000000"),
@@ -416,10 +423,6 @@ def test_distribution():
             1,
         )
     ]
-    assert (
-        get_distribution(REPO, "00000", case_id, context_id, machine_hash, 10).all()
-        == []
-    )
 
     # ----- set_z_scores
 
@@ -500,13 +503,14 @@ def test_distribution_multiple_runs_same_commit():
     machine_hash = summary_1.run.machine.hash
 
     assert get_distribution(
-        REPO, "xxxxx", case_id, context_id, machine_hash, 10
+        REPO, "xxxxx", case_id, context_id, commit_1.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "xxxxx",
             case_id,
             context_id,
+            commit_1.id,
             MACHINE,
             "s",
             decimal.Decimal("2.0000000000000000"),
@@ -524,13 +528,14 @@ def test_distribution_multiple_runs_same_commit():
     ]
 
     assert get_distribution(
-        REPO, "yyyyy", case_id, context_id, machine_hash, 10
+        REPO, "yyyyy", case_id, context_id, commit_2.id, machine_hash, 10
     ).all() == [
         (
             REPO,
             "yyyyy",
             case_id,
             context_id,
+            commit_2.id,
             MACHINE,
             "s",
             decimal.Decimal("2.5000000000000000"),

migrations/versions/dc0ed346df63_add_commit_id_to_dist_table.py

@@ -0,0 +1,34 @@
+"""Add commit id to dist table
+
+Revision ID: dc0ed346df63
+Revises: 69493ddc938a
+Create Date: 2021-07-28 08:50:54.190008
+
+"""
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision = "dc0ed346df63"
+down_revision = "69493ddc938a"
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    op.add_column(
+        "distribution", sa.Column("commit_id", sa.String(length=50), nullable=True)
+    )
+    op.create_index(
+        "distribution_commit_id_index", "distribution", ["commit_id"], unique=False
+    )
+    op.create_foreign_key(
+        None, "distribution", "commit", ["commit_id"], ["id"], ondelete="CASCADE"
+    )
+
+
+def downgrade():
+    op.drop_constraint(None, "distribution", type_="foreignkey")
+    op.drop_index("distribution_commit_id_index", table_name="distribution")
+    op.drop_column("distribution", "commit_id")