--- a/bench.py	Thu May 07 13:21:40 2015 +0800
+++ b/bench.py	Thu May 07 13:23:14 2015 +0800
@@ -25,7 +25,9 @@
 
 
 parser = ArgumentParser(description='Benchmark revisions and put results in the db.')
-parser.add_argument('revsets', metavar='REVSET', default=('last(all(), 120)',), nargs='*', help='update these revisions')
+group = parser.add_mutually_exclusive_group()
+group.add_argument('revsets', metavar='REVSET', default=('last(all(), 120)',), nargs='*', help='update these revisions')
+group.add_argument('--auto', metavar='MAXREVS', type=int, help='guess revisions, up to MAXREVS')
 parser.add_argument('--retry', action='store_true', help='try and reduce existing timings')
 
 
@@ -37,6 +39,75 @@
     return output.split()
 
 
+def guessnew(maxrevs):
+    """ Pick one continuous span of nodes that still need testing. """
+    cmd = [HG, 'log', '-R', TESTHGREPO, '-T', '{node}\n', '-r', 'sort(all(), rev)']
+    output = subprocess.check_output(cmd)
+    nodes = output.split()
+    picking = False
+    todo = []
+    conn = sqlite3.connect(DBPATH)
+    while len(todo) < maxrevs:
+        node = nodes.pop()
+        count = conn.execute('SELECT COUNT(*) FROM results WHERE node = ?', (node,)).fetchone()[0]
+        if count < len(MARKS) * len(('without cache', 'with cache')):
+            todo.append(node)
+            picking = True
+        elif picking:
+            break
+    conn.close()
+    return todo
+
+
+def guessspikes(maxrevs):
+    cmd = [HG, 'log', '-R', TESTHGREPO, '-T', '{node}\n', '-r', 'sort(all(), -rev)']
+    output = subprocess.check_output(cmd)
+    nodes = output.split()
+    todo = []
+    conn = sqlite3.connect(DBPATH)
+    limits = {
+        mark:
+            conn.execute(
+                'SELECT MIN(time), MAX(time) FROM results WHERE mark = ? AND cache = ?',
+                (mark, False)).fetchone()
+            +
+            conn.execute(
+                'SELECT MIN(time), MAX(time) FROM results WHERE mark = ? AND cache = ?',
+                (mark, True)).fetchone()
+        for mark in MARKS
+    }
+    results = {}
+    for node in nodes:
+        resultsq = conn.execute(
+            'SELECT mark, time, cache FROM results WHERE node = ?',
+            (node,))
+        for mark, t, cache in resultsq:
+            results.setdefault(node, {}).setdefault(mark, [None, None])
+            if not cache:
+                results[node][mark][0] = t
+            else:
+                results[node][mark][1] = t
+    conn.close()
+    for i in range(1, len(nodes) - 1):
+        if len(todo) >= maxrevs:
+            break
+        node1 = nodes[i - 1]
+        node2 = nodes[i]
+        node3 = nodes[i + 1]
+        for mark in MARKS:
+            for cache in (False, True):
+                try:
+                    eps = abs(results[node1][mark][cache] - results[node3][mark][cache])
+                    delta = results[node2][mark][cache] - results[node1][mark][cache]
+                    l = limits[mark][2:4] if cache else limits[mark][0:2]
+                    if delta > eps * 10 and delta > (l[1] - l[0]) * 0.1:
+                        if node2 not in todo:
+                            todo.append(node2)
+                except (KeyError, TypeError):
+                    continue
+    return todo
+
+
 def test(mark, mintime=1.0, mintries=3, dropcache=True):
     results = []
 
@@ -156,11 +227,16 @@
     os.remove(LOCKFILE)
 
 
-def main(revsets, retry=False):
+def main(args):
     lock()
     try:
         dbinit()
-        dbupdate(revsets, retry)
+        if args.auto:
+            if args.retry:
+                args.revsets = guessspikes(args.auto)
+            else:
+                args.revsets = guessnew(args.auto)
+        dbupdate(args.revsets, args.retry)
     finally:
         unlock()
 
@@ -168,4 +244,4 @@
 if __name__ == '__main__':
     logging.getLogger().setLevel(logging.INFO)
     args = parser.parse_args()
-    main(args.revsets, args.retry)
+    main(args)