nheight

xmetrics/Dockerfile

@@ -47,7 +47,7 @@ RUN git fetch && git checkout f81543a34ee363dcc00e8632fd7cfcd4a3478b23
 ENV SRCDIR="."
 ENV SRCPATTERN="*.[ach]*"
 
-FROM metrics-radn as metrics-repo
+FROM metrics-cargo as metrics-repo
 
 FROM metrics-repo as metrics-commits
 COPY common.py /code/common.py

xmetrics/commits.py

@@ -4,22 +4,33 @@ from subprocess import check_output
 
 from common import counter
 
+BITS = 7
+
+def mask(height: int) -> int:
+    return ((1 << height.bit_length()) - 1) >> BITS
+
+def valid(commit: str, height: int) -> bool:
+    expected = int.from_bytes(bytes.fromhex(commit), "little")
+    return mask(height) & (height ^ expected) == 0
+
 args = []
 original_commits = check_output(["git", "log", "--pretty=%H", *args], text=True).splitlines()
 N = len(original_commits)
-BITS = 10
 total_changes: dict[str, tuple[int, int]] = {}
 chosen_parents: dict[str, str | None] = {}
 heights: dict[str, int] = {}
+nheights: dict[str, int] = {}
 included = set()
 for i, commit in enumerate(reversed(original_commits)):
     print(f"C={i / N:6f}", flush=True)
     parents = check_output(["git", "log", "--pretty=%P", "-n", "1", commit], text=True).split()
+    parents = [parent for parent in parents if parent in heights]
     chosen_parent = None
     height = 1 + max((heights[parent] for parent in parents), default=0)
-    mask = ((1 << height.bit_length()) - 1) >> BITS
-    expected = int.from_bytes(bytes.fromhex(commit), "little")
-    if mask & (height ^ expected) == 0:
+    nheight = 1 + min((nheights[parent] for parent in parents), default=0)
+    if valid(commit, height) or nheight > mask(height):
+        included.add(commit)
+        nheight = 0
         ctr = counter(commit, True)
         changes = ctr.total(), 1
         for parent in parents:
@@ -36,7 +47,6 @@ for i, commit in enumerate(reversed(original_commits)):
             if maybe_changes > changes:
                 changes = maybe_changes
                 chosen_parent = parent
-        included.add(commit)
     else:
         changes = 0, 1
         for parent in parents:
@@ -51,6 +61,7 @@ for i, commit in enumerate(reversed(original_commits)):
     assert chosen_parent is None or chosen_parent in included
     chosen_parents[commit] = chosen_parent
     heights[commit] = height
+    nheights[commit] = nheight
 commit = original_commits[0]
 commits = []
 while commit is not None:

xmetrics/metrics.py

@@ -13,7 +13,7 @@ C = min(len(commits), 720)
 for i, commit in enumerate(reversed(commits)):
     print(f"P={i / len(commits):6f}", flush=True)
     print("running", commit, flush=True)
-    current_ctr = counter(commit, True)
+    current_ctr = counter(commit, False)
     added = current_ctr - last_ctr
     deleted = last_ctr - current_ctr
     entries.append((i, current_ctr.total(), added.total(), deleted.total()))