[vwnc] VW7.6 -- HELP I have a class that is acting in a very strange manner

Terry

===========================================================
Terry Raymond
Crafted Smalltalk
80 Lazywood Ln.
Tiverton, RI  02878
(401) 624-4517      [hidden email]
<http://www.craftedsmalltalk.com>
===========================================================

Thank you to everyone -- all the thoughts and suggestions helped me to think
about what might be happening.  Here is a quick summary, although I may never
know exactly the events that caused the exact symptoms.

1. turns out it was NOT just this class, this was a little used table view (DataSet actually)
    of a business object.
2. the methods were those accessed by an AspectAdaptor connected to the DataSet.
3. latest symptom, a breakpoint in the method, use Event-Faithful Debugging,
    the break would occur at the END of the method (a number of statements past the
    breakpoint)
4. seems that in my window-event-handler, for any window event, I was updating the
    valueHolder for the object of the DataSet and thus re-kicking all the AspectAdaptor's.
    This could get a bit much for anyone -- window events do occur quite often when
    you do something!
    Turns out it should only have kicked the valueHolder IF it also caused a
    popup window to close (there was no popup window) -- it was just sloppy code.

I certainly would expect the above to cause recursions on exceptions, and UI stuff is
often difficult to debug anyway, so fixing this fixed my problem!!

Reinout Heeck wrote:

Here's some buckshot:

-The debugger will recurse if you put breakpoints in code that the  

debugger needs,

this happens when putting breakpoints in the View hierarchy for  

instance.

In your case: is any of your code registered with/used by the debugger  

in any way, like would happen if you implemented your own  

#inspectorExtraAttributes (or perhaps merely #printString) on your  

object. Is #printString exercising either of your two probed methods?

-The debugger will hang the system if you step into some critical code  

(that is protected by a semaphore) that the debugger or rest of the  

system needs. The lock around symbol table allocations might be an  

example.

-To avoid these kind of locks the system contains a sprinkling of  

pixiedust (see the senders of #valueUninterruptably). This method  

might explain your debugger opening at the second breakpoint/halt  

instead of the first.

-Finally: halts and breakpoints are mere exceptions. If you have an  

exception handler that is trying to do something 'smart' with all  

exception types it might break the debugger - but you made explicit  

that this is not the case...

HTH,

Reinout

-------

On Feb 25, 2009, at 9:56 PM, Dennis Smith wrote:

I have a class with a number if ivars (no conflicts, I checked)

and quite a few methods, mostly read and write accessors for the ivars

with behavior, not just accessing.

I place a halt at the start of one of the methods and also in a method

that this method invokes (in another object/class).

One expects it to stop on the first halt (I have no exception  

handlers,

just the default VW handler) and then on proceed to stop at the next  

halt.

What occurs is that it stops at the second halt, showing in the stack

the first method with a halt and other methods prior to it which  

appear

correct.

If I click on something else on the stack in the debugger, nothing  

happens,

still looking at the 2nd halt.  Proceed in the debugger freezes the  

image.

If I try and do pretty much the same thing with any other class, no  

problem

(we have lots of business classes subclassed from the same main  

business

   class which have similar structure).

I looked for methods in the offending class that might conflict with

something

in Object, ran a script to check -- there are none, instance or  

class side.

Anyone have any thoughts on what kind of thing I might look for?

-- 

Dennis Smith                                    +1 416.798.7948

Cherniak Software Development Corporation   Fax: +1 416.798.0948

509-2001 Sheppard Avenue East        [hidden email]

Toronto, ON M2J 4Z8              <a
href="sip:dennis@CherniakSoftware.com">sip:dennis@...

Canada                           http://www.CherniakSoftware.com

Entrance off Yorkland Blvd south of Sheppard Ave east of the DVP

_______________________________________________

vwnc mailing list

[hidden email]

http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

_______________________________________________

vwnc mailing list

[hidden email]

http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

-- 

Dennis Smith                                  +1 416.798.7948

Cherniak Software Development Corporation   Fax: +1 416.798.0948

509-2001 Sheppard Avenue East        [hidden email]

Toronto, ON M2J 4Z8              <a
href="sip:dennis@CherniakSoftware.com">sip:dennis@...

Canada                         http://www.CherniakSoftware.com

Entrance off Yorkland Blvd south of Sheppard Ave east of the DVP