>>You're familiar with the F8 key? At times like this, I'm not sure there's anything
>>you can do but track down the exact location of failure. Usually that will lend
>>*some* hint.
The problem is that it seems to be a moving target. The act of stepping
through code seems to move the point of failure.
I have actually gone back to the much older technique of putting in trace
points and tracing to file. It seems to break at bizarre times. For
example, it can go through the trace point at the end of the subroutine but
fail to get back to the calling code.
Sometimes it doesn't break in the application at all. It's the IDE which
breaks - usually just after terminating a debug session.
I have finally given up. I have reverted to the last working version (thank
heavens for Visual Source Safe!) which does not exhibit the problem. I
shall now add my new feature piece by piece, watching for any signs of it
playing up. Fortunately, the delays have enabled me to think of a way to
achieve what I was trying to do in a slightly simpler way.
Keith
This sounds like a memory cleanup error. Usually, it has something to
do with string pointers and API calls and is a manifestation of
releasing memory allocated to a string that has already been released.
Were you making any API calls that involved string variables in the
code for your feature? If not, it could be a problem in one of the
components you are using.
HTH,
Bryan
____________________________________________________________
New Vision Software "When the going gets weird,"
Bryan Stafford "the weird turn pro."
alpine_don'tsendspam@mvps.org Hunter S. Thompson -
Microsoft MVP-Visual Basic Fear and Loathing in LasVegas
>>>You're familiar with the F8 key? At times like this, I'm not sure there's
>anything
[quoted text clipped - 19 lines]
>
>Keith
Keith Sheppard - 27 May 2005 09:13 GMT
>>Were you making any API calls that involved string variables in the
>>code for your feature?
It is entirely possible. Current status is I have scrapped the enhancements
and am re-adding them a bit at a time using a slightly different design
philosophy. I shall look out for this as an area to be wary of.
Something good came out of the first attempt in that I can avoid a few blind
alleys second time around.
Keith
Keith Sheppard - 27 May 2005 09:15 GMT
>>This sounds like a memory cleanup error.
Another possibility which occurred to me is that I had a number of instances
where a variable of type Object could have contained a pointer to a number
of different object types depending upon context. I did wonder whether VB
could have got confused and called the destructor for the wrong object type
when it finally did its cleanup.
Keith