[Mono-bugs] [Bug 41989][Wis] Changed - Volatile emit/use missing

bugzilla-daemon@rocky.ximian.com bugzilla-daemon@rocky.ximian.com
Mon, 28 Apr 2003 08:07:33 -0400 (EDT)

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@ximian.com.


--- shadow/41989	Sun Apr 27 13:24:31 2003
+++ shadow/41989.tmp.21029	Mon Apr 28 08:07:33 2003
@@ -1,14 +1,14 @@
 Bug#: 41989
 Product: Mono/MCS
 Version: unspecified
+OS: unknown
 OS Details: 
 Status: NEW   
+Severity: Unknown
 Priority: Wishlist
 Component: Misc
 AssignedTo: mono-bugs@ximian.com                            
 ReportedBy: miguel@ximian.com               
 QAContact: mono-bugs@ximian.com
 TargetMilestone: ---
@@ -24,6 +24,21 @@
 .field reqd [mscorlib]System.Runtime.CompilerServices.IsVolatile int32 a
 Maybe we could store this information in the FieldAttributes (using an
 unused bit there) for the time being, because using a dynamic invocation
 for "IsVolatile" might be very expensive.
+------- Additional Comments From lupus@ximian.com  2003-04-28 08:07 -------
+The important thing when doing a dynamic call is to get the MethodInfo
+just once. For example, in a test case I wrote my !.13 GHz cpu can do
+10000 ctor invocations in 50 ms. 
+I assume you need to do the call for each field acces: both corlib and
+mcs have about 11000 instructions to access fields: in this case the
+overhead in the compilation time would be less than 1 % doing the
+dynamic invocation.
+There is also one other thing to consider: it is very rare to access a
+field in some referenced assembly, so most of the time, mcs has all
+the info it needs in ModFlags and it doesn't need to do a dynamic call
+at all (it seems to do that already).
+So, since the operation turns out to be rare anyway, I'd go with the
+dynamic call. Let me know if my assumptions are wrong.