Somewhere in the mailing list archives are notes about a VM I built  
that writes each message send (plus other data) to a text file.
This data truly represents which messages the VM is sending.

mmm look on my idisk via  http://www.smalltalkconsulting.com/squeak.html

in the Experimental/MessageLoggingVM  folder

Tools to interpret the trace data would be helpful.


On Jan 27, 2008, at 9:04 PM, Matthew Fulmer wrote:

> On Sat, Jan 26, 2008 at 10:07:03PM -0800, Jerome Peace wrote:
>> This is mostly external observations. At the time I
>> reported this I hadn't learned to use the tally tools
>> usefully. (I still haven't. They give too much
>> information. I want something that will point me
>> quickly to the culprit.)
>
> I talked to Bryce about this on #squeak, and he gave several
> suggestions:
>
> - In MessageTally reports, look for big drops in time usage from
>  a parent to child method. That usually points toward the
>  culprit
> - Insert logging statements in code. In Keith's Logging
>  framework (http://wiki.squeak.org/squeak/3706), I added a
>  method just to make logging profiles easy. It helps to see
>  when a method is getting called way too often.
>
> I still don't see how to use the debugger effectively to profile
> UI's, since events will be posted from the debugger thread, but
> screen updates happen in the UI thread, which is impossible to
> get a useful debugger on (as far as I can tell)
>
> -- 
> Matthew Fulmer -- http://mtfulmer.wordpress.com/
> Help improve Squeak Documentation: http://wiki.squeak.org/squeak/808
>

--
= 
= 
= 
========================================================================
John M. McIntosh <johnmci at smalltalkconsulting.com>
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
= 
= 
= 
========================================================================