[IronPython] ScriptScope.SetVariable() performance
Dino Viehland
dinov at microsoft.com
Thu Aug 12 19:24:59 CEST 2010
Oh, yeah, objects aren't dynamic unless they're typed to dynamic :) I guess it needs to be:
((dynamic)mScope).PH = editScope;
From: users-bounces at lists.ironpython.com [mailto:users-bounces at lists.ironpython.com] On Behalf Of Cory Brostowicz
Sent: Thursday, August 12, 2010 10:11 AM
To: Discussion of IronPython
Subject: Re: [IronPython] ScriptScope.SetVariable() performance
I'm not sure what you mean by "mScope.PH = editScope". mScope = ScriptScope and doesn't allow dynamic properties.
On the other hand, thanks for the suggestion with the object cast. That gave brought my setup performance up to par with the native C# edits!
Edit: SetProviderType Count: 70291 Min: 9.866000E-001 Max: 5.393440E+001 Total: 6.684688E+004 Setup: 1.904770E+002
Edit: SetFeeSchedule Count: 70291 Min: 1.229300E+000 Max: 6.506160E+001 Total: 9.587090E+004 Setup: 1.747695E+002
Edit: phDefaultNetworkCode Count: 70291 Min: 9.700000E-003 Max: 2.347088E+002 Total: 1.154772E+003 Setup: 1.156382E+002
Edit: phCloneFieldList Count: 70291 Min: 3.200000E-003 Max: 4.007700E+000 Total: 2.727729E+002 Setup: 1.137180E+002
Edit: phAssignValueToField Count: 70291 Min: 4.400000E-003 Max: 1.205800E+000 Total: 4.475063E+002 Setup: 1.152221E+002
On Thu, Aug 12, 2010 at 11:42 AM, Dino Viehland <dinov at microsoft.com<mailto:dinov at microsoft.com>> wrote:
I'd suggest doing mScope.PH = editScope. That will create a rule which will get cached and will generally run faster than passing the string directly.
The only other thing I can think of is maybe there's something odd w/ rule being produced because "editScope" is a dynamic object in SetupEdit so "mScope.SetVariable("PH", editScope);" ends up dispatching dynamically. If you did "mScope.SetVariable("PH", (object)editScope);" it would dispatch statically.
If neither of those solutions helps I can see if I can look deeper and see if I can repro it. There were some issues w/ IronPython 2.6 where this was really slow but they should have been fixed w/ 2.6.1.
From: users-bounces at lists.ironpython.com<mailto:users-bounces at lists.ironpython.com> [mailto:users-bounces at lists.ironpython.com<mailto:users-bounces at lists.ironpython.com>] On Behalf Of Cory Brostowicz
Sent: Thursday, August 12, 2010 7:34 AM
To: users at lists.ironpython.com<mailto:users at lists.ironpython.com>
Subject: [IronPython] ScriptScope.SetVariable() performance
Hello,
I'm trying to get more performance out of my IronPython scripts inside one of my applications, and noticed I'm spending the bulk of my time setting up the ScriptScope prior to execution. My applications translates information coming from a flat file, and I use IronPython to enforce business rules on various records that I pull from my input file. The object that I'm passing to the scope is custom implementation of the ExpandoObject.
Am I doing anything out of the ordinary here, or is there a better way to get the performance I'm after? It doesn't seem like setting up the script should performance significantly slower than executing the script itself.
Here is an example of the performance results I'm getting. I'm keeping track of various timings for execution time, but the script setup timer is really just stopping and starting a stopwatch before and after each call to ScriptScope.SetVariable()...
DUMPING EDIT STATS
------------------
Edit: SetProviderType Count: 70291 Min: 9.927000E-001 Max: 4.922160E+001 Total: 7.656987E+004 Setup: 4.667084E+005
Edit: SetFeeSchedule Count: 70291 Min: 1.572500E+000 Max: 6.547200E+001 Total: 1.067754E+005 Setup: 4.639014E+005
Edit: phDefaultNetworkCode Count: 70291 Min: 1.900000E-002 Max: 2.413129E+002 Total: 1.618541E+003 Setup: 1.161743E+002
Edit: phCloneFieldList Count: 70291 Min: 4.400000E-003 Max: 4.083400E+000 Total: 3.565259E+002 Setup: 1.113712E+002
Edit: phAssignValueToField Count: 70291 Min: 9.700000E-003 Max: 1.777100E+000 Total: 5.765196E+002 Setup: 1.108866E+002
SetProviderType and SetFeeSchedule are both IronPython scripts, the other three are C# edits that I'm passing my ExpandoObject directly to.
4.667 E+005 = almost 8 minutes...
1.108 E+002 = .1 seconds
Thanks in advance for any advice you can help me with.
-Cory
class IronPythonScriptHost : PrimeEditBase {
ScriptSource mScriptSource;
ScriptEngine mEngine;
ScriptScope mScope;
public IronPythonScriptHost(string scriptText, string ScriptName, CommandLine mCommandLine, string[] mEditParms, ScriptEngine engine)
: base(mCommandLine, mEditParms) {
mEditName = ScriptName;
mIsScript = true;
mEngine = engine;
mScriptSource = mEngine.CreateScriptSourceFromString(scriptText, Microsoft.Scripting.SourceCodeKind.File);
mScriptSource.Compile();
mScope = mEngine.CreateScope();
}
public override void SetupEdit(dynamic editScope) {
StartSetupTimer();
mScope.SetVariable("PH", editScope);
StopSetupTimer();
}
public override void performEdit() {
StartExecutionTimer();
try {
mScriptSource.Execute(mScope);
CleanupEdit();
} catch (Exception ex) {
ExceptionOperations ExcOps = mEngine.GetService<ExceptionOperations>();
throw new Exception("Exception during execution of " + base.EditName + ". " + ex.Message + "\r\n" + ExcOps.FormatException(ex));
}
StopAndRecordExecutionTimer();
}
}
_______________________________________________
Users mailing list
Users at lists.ironpython.com<mailto:Users at lists.ironpython.com>
http://lists.ironpython.com/listinfo.cgi/users-ironpython.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ironpython-users/attachments/20100812/53eccb30/attachment.html>
More information about the Ironpython-users
mailing list