Random Crash Using Infragistics UltraWinToolbars

[Charles Law]

I seem to be posing a lot at the moment, so please accept my apologies.

I can't say, categorically that the Infragistics tools are to blame, but I
have had this intermittent problem now for two months, and all attempts to
localise it have failed. I have looked at the Infragistics newsgroups and
web sites, and there are hints that there are potential problems in this
area, especially in a multi-threaded app (I have two or three threads
running).

Sometimes I can get the crash to be repeatable, but just as I home in on it,
it stops happening. Then, some time later it will happen again in slightly
different circumstances.

The application terminates without warning, and without executing any
unhandled exception handlers. In Debug | Exceptions, I have all four - C++
Exceptions, CLR Exceptions, Native Run-Time Checks, and Win32 Exceptions -
set to break into the debugger when the exception is thrown and when it is
not handled, but no exception is thrown. The output window shows that the
app terminated with exit code 0.

The only common element seems to be that the crash happens as I click, or
immediately after I click a menu item or toolbar button (both Infragistics
tools).

My reason for posting is just in case anyone else has had similar
experience, or whether anyone can suggest a way of identifying what the app
was doing just before the crash. As it does not happen in a common place in
my code, I am assuming that there must be a common place in the Infragistics
code, but I don't know how to find or prove it.

Thanks in advance for any pointers.

Charles

[Charles Law]

> I seem to be posing a lot at the moment, so please accept my apologies.

... posting ...

Anyway, in case anyone is watching this post with keen anticipation, I think
I have found the problem.

It's all to do with overlapped file i/o and the fact that sometimes, between
waiting for single object and getting the final overlapped result,
Windows/.NET/whatever can move stuff about. So, what you were pointing to at
the start you might not be pointing to at the end.

So now I marshal my overlapped structure in and out of the calls - like I
should have done in the first place - and it hasn't crashed since; I just
know I'm going to eat those words.

Infragistics, you are reprieved (for now).

Charles
"Charles Law" <bl***@nowhere.com> wrote in message
news:uM*************@TK2MSFTNGP09.phx.gbl...

I seem to be posing a lot at the moment, so please accept my apologies.

I can't say, categorically that the Infragistics tools are to blame, but I
have had this intermittent problem now for two months, and all attempts to
localise it have failed. I have looked at the Infragistics newsgroups and
web sites, and there are hints that there are potential problems in this
area, especially in a multi-threaded app (I have two or three threads
running).

Sometimes I can get the crash to be repeatable, but just as I home in on it, it stops happening. Then, some time later it will happen again in slightly
different circumstances.

The application terminates without warning, and without executing any
unhandled exception handlers. In Debug | Exceptions, I have all four - C++
Exceptions, CLR Exceptions, Native Run-Time Checks, and Win32 Exceptions -
set to break into the debugger when the exception is thrown and when it is
not handled, but no exception is thrown. The output window shows that the
app terminated with exit code 0.

The only common element seems to be that the crash happens as I click, or
immediately after I click a menu item or toolbar button (both Infragistics
tools).

My reason for posting is just in case anyone else has had similar
experience, or whether anyone can suggest a way of identifying what the app was doing just before the crash. As it does not happen in a common place in my code, I am assuming that there must be a common place in the Infragistics code, but I don't know how to find or prove it.

Thanks in advance for any pointers.

Charles