[Mono-bugs] [Bug 80508][Nor] Changed - code quality regression for array init

[bugzilla-daemon at bugzilla.ximian.com]

Please do not reply to this email- if you want to comment on the bug, go to the
URL shown below and enter your comments there.

Changed by lupus at ximian.com.

http://bugzilla.ximian.com/show_bug.cgi?id=80508

--- shadow/80508	2007-01-12 19:13:33.000000000 -0500
+++ shadow/80508.tmp.21500	2007-01-13 08:50:08.000000000 -0500
@@ -50,6 +50,19 @@
 I understand that on other architectures you can get different results
  so it probably makes better sense to optimize InitializeArray to be
 architecture specific and set threshold lower to cover only cost of
 managed->unmanaged transition.
 
 
+
+------- Additional Comments From lupus at ximian.com  2007-01-13 08:50 -------
+I'm not saying that currently the performance would be better. I'm
+saying that this change causes orrible code to be persisted in
+assemblies and that code is not optimizable. Changes like this should
+be done only after discussion and consensus, because the generated
+code is persisted and can live forever. You can run your benchmarks
+all you want, but you're measuring the wrong thing.
+This change increases jit time and runtime memory use and it's
+preventing the JIT from optimizing the initialization (the jit could
+just mempcy() the data on little endian systems).
+This has nothing to do with different architectures: the issue is that
+this change is not an optimization: it is a pessimization.