Rev 4: Force all of the texts to have completely unique strings before doing the comparison. in http://bzr.arbash-meinel.com/plugins/bench_patience

[John Arbash Meinel]

At http://bzr.arbash-meinel.com/plugins/bench_patience

------------------------------------------------------------
revno: 4
revision-id:john at arbash-meinel.com-20071205001636-9fvdex40n983rz3w
parent: john at arbash-meinel.com-20071204235627-07ldd8e3pk51jo15
committer: John Arbash Meinel <john at arbash-meinel.com>
branch nick: bench_patience
timestamp: Tue 2007-12-04 18:16:36 -0600
message:
  Force all of the texts to have completely unique strings before doing the comparison.
modified:
  __init__.py                    __init__.py-20071204154123-hrdofqdugj0iwxcy-1
-------------- next part --------------
=== modified file '__init__.py'
--- a/__init__.py	2007-12-04 23:56:27 +0000
+++ b/__init__.py	2007-12-05 00:16:36 +0000
@@ -43,31 +43,32 @@
             texts = w.get_line_list(versions)
             total_len = sum(len(t) for t in texts)
             self.outf.write('%d total bytes\n' % total_len)
-            last_text = None
+            last_text = []
             from bzrlib.patiencediff import PatienceSequenceMatcher
             times = []
             pb = ui.ui_factory.nested_progress_bar()
             try:
                 t_overall_start = time.clock()
+                total_time = 0
                 for idx, text in enumerate(texts):
+                    # Make sure all the strings are brand new
+                    text = [l[:10] + l[10:] for l in text]
                     pb.update('comparing', idx, len(texts))
-                    if last_text is None:
-                        last_text = text
-                        continue
+                    t_start = time.clock()
                     matcher = PatienceSequenceMatcher(None, last_text, text)
-                    last_text = None
-                    t_start = time.clock()
                     matcher.get_matching_blocks()
                     t_delta = time.clock() - t_start
                     times.append(t_delta)
+                    last_text = [l[:10] + l[10:] for l in text]
                 total_time = time.clock() - t_overall_start
             finally:
                 pb.finished()
-            avg = sum(times) / float(len(times))
+            sum_time = sum(times)
+            avg = sum_time / float(len(times))
             max_time = max(times)
             min_time = min(times)
-            print 'Total: %.3fs, min: %.3fms avg: %.3fms max: %.3fms' % (
-                total_time, min_time*1000, avg*1000, max_time*1000)
+            print 'Total: %.3fs/%.3fs, min: %.3fms avg: %.3fms max: %.3fms' % (
+                total_time, sum_time, min_time*1000, avg*1000, max_time*1000)
         finally:
             tree.unlock()