fix(benchmarks): address review feedback on workload comparisons and counts

<!-- agent -->
Address benchmark review feedback on the generated fixtures. Copilot noted
that "performance tests still report their iteration limits of 1,000–10,000
as processed counts" even though the fixture has only 90 packed objects.
Use the existing streaming counters and clamp list-based limits to the
available SHA count so both totals and throughput reflect completed work.

Copilot also requested: "Include the benchmark harness sources in the
fingerprint, and add a harness-only change to the digest regression test."
Hash the Python sources in `test/performance` alongside `VERSION` and `git`.
This prevents comparing different timed workloads that return the same
result. Extend the generated source fixture and document the digest scope.

Assisted-by: GPT 6.0 Astra
Co-authored-by: GPT 6.0 Astra <codex@openai.com>

Author: Byron

gitdb/gitdb/test/performance/test_pack.py

@@ -51,7 +51,7 @@ def test_pack_random_access(self):
         # END for each random mode
 
         # query info and streams only
-        max_items = 10000           # can wait longer when testing memory
+        max_items = min(10000, ns)  # can wait longer when testing memory
         for pdb_fun in (pdb.info, pdb.stream):
             st = time()
             for sha in sha_list[:max_items]:
@@ -62,7 +62,7 @@ def test_pack_random_access(self):
         # END for each function
 
         # retrieve stream and read all
-        max_items = 5000
+        max_items = min(5000, ns)
         pdb_stream = pdb.stream
         total_size = 0
         st = time()

gitdb/gitdb/test/performance/test_pack_streaming.py

@@ -49,7 +49,7 @@ def test_pack_writing(self):
         # END gather objects for pack-writing
         elapsed = time() - st
         print("PDB Streaming: Got %i streams by sha in in %f s ( %f streams/s )" %
-              (ni, elapsed, ni / (elapsed or 1)), file=sys.stderr)
+              (count, elapsed, count / (elapsed or 1)), file=sys.stderr)
 
         st = time()
         PackEntity.write_pack((pdb.stream(sha) for sha in pdb.sha_iter()), ostream.write, object_count=pdb.size())
@@ -78,4 +78,4 @@ def test_stream_reading(self):
         elapsed = time() - st
         total_kib = total_size / 1000
         print(sys.stderr, "PDB Streaming: Got %i streams by sha and read all bytes totallying %i KiB ( %f KiB / s ) in %f s ( %f streams/s )" %
-              (ni, total_kib, total_kib / (elapsed or 1), elapsed, ni / (elapsed or 1)), sys.stderr)
+              (count, total_kib, total_kib / (elapsed or 1), elapsed, count / (elapsed or 1)), sys.stderr)

test/performance/README.md

@@ -99,6 +99,7 @@ The CI job publishes all means, standard deviations and ratios, retains raw
 artifacts, and uses `pyperf compare_to` for significance reporting. It fails on
 execution/parity errors, without a noisy speed threshold. Compare like-for-like
 fixture revisions and source content digests when tracking performance across backend changes.
-The source digest covers `VERSION` and the Python package files, including local
-edits. Measurements from the old GitPython-history fixture are not comparable
+The source digest covers `VERSION`, the Python package files, and the benchmark
+harness sources in `test/performance`, including local edits.
+Measurements from the old GitPython-history fixture are not comparable
 with this generated workload.

test/performance/bench_repository.py

@@ -126,7 +126,8 @@ def worker_arguments(cmd, args):
 
 def source_digest(root):
     digest = sha256()
-    for path in sorted([root / "VERSION", *(root / "git").rglob("*.py")]):
+    sources = [root / "VERSION", *(root / "git").rglob("*.py"), *(root / "test" / "performance").glob("*.py")]
+    for path in sorted(sources):
         digest.update(path.relative_to(root).as_posix().encode() + b"\0" + path.read_bytes() + b"\0")
     return digest.hexdigest()
 

test/performance/test_benchmark_comparison.py

@@ -29,4 +29,5 @@ def test_source_digest_detects_code_changes_without_git_metadata():
 
     fixture = read_only_fixture("benchmark-source.sh")
     assert source_digest(fixture / "source") != source_digest(fixture / "copy")
+    assert source_digest(fixture / "source") != source_digest(fixture / "harness")
     assert source_digest(fixture / "source") == source_digest(fixture / "moved")

tests/fixtures/benchmark-source.sh

@@ -1,8 +1,11 @@
 #!/usr/bin/env bash
 set -eu
-mkdir -p source/git
+mkdir -p source/git source/test/performance
 printf '1.0\n' > source/VERSION
 printf 'value = 1\n' > source/git/module.py
+printf 'value = 1\n' > source/test/performance/bench_repository.py
 cp -R source copy
 cp -R source moved
+cp -R source harness
 printf 'value = 2\n' > copy/git/module.py
+printf 'value = 2\n' > harness/test/performance/bench_repository.py