[llvm-commits] [zorg] r153973 - /zorg/trunk/lnt/lnt/db/runinfo.py

[Daniel Dunbar]

Author: ddunbar
Date: Tue Apr  3 16:56:29 2012
New Revision: 153973

URL: http://llvm.org/viewvc/llvm-project?rev=153973&view=rev
Log:
[LNT] When computing a run comparison without any std.dev. (estimated or
otherwise), use a minimum limit of 0.2% instead of the historical 5%.
 - These days, this situation only commonly occurs when individual users are
   looking at LNT results (since any real server will have enough data to
   estimate the std.dev.), and in such cases users generally want to see almost
   all the information.

Modified:
    zorg/trunk/lnt/lnt/db/runinfo.py

Modified: zorg/trunk/lnt/lnt/db/runinfo.py
URL: http://llvm.org/viewvc/llvm-project/zorg/trunk/lnt/lnt/db/runinfo.py?rev=153973&r1=153972&r2=153973&view=diff
==============================================================================
--- zorg/trunk/lnt/lnt/db/runinfo.py (original)
+++ zorg/trunk/lnt/lnt/db/runinfo.py Tue Apr  3 16:56:29 2012
@@ -96,8 +96,10 @@
             else:
                 return UNCHANGED_PASS
 
-        # Otherwise, use the old "significant change" metric of > 5%.
-        if abs(self.pct_delta) >= .05:
+        # Otherwise, report any changes above 0.2%, which is a rough
+        # approximation for the smallest change we expect "could" be measured
+        # accurately.
+        if abs(self.pct_delta) >= .002:
             if self.pct_delta < 0:
                 return IMPROVED
             else: