TamarGranor
Programmer
An app I've been working on that's nearly ready for beta testing is giving random C5 errors and I'm looking for ideas.
Here's the background: I have a VFP EXE (call it UI) that instantiates a VFP out-of-process COM object (call it DL). DL instantiates another VFP COM object (call it OnlineTimer) as well as a C# COM object (call it Comms).
The C# object, Comms, talks to some hardware and passes information between the hardware and DL. When it has information ready for DL, it sets a flag. DL has a timer that interrogates that flag and when there's data available, processes it. Processing the data means parsing the XML format that Comms returns and, generally, poking updates into the object model maintained by UI. Many, many lines of code in DL look something like:
This.oCallingApp.oNetwork.DoSomething()
They mostly drill more layers down into the object model before actually calling a method.
After DL finishes processing one returned item, it makes a call to a UI method that repaints the interface. Each form has a method that can be called to update its display with the latest data.
The C5s are happening in DL. Based on the VFP9Rerr.LOG file that's automatically created, they happen in various places, but it looks like the line that crashes is always one that includes This.oCalling.something. (I don't have enough data yet to be sure that's the case.)
Naturally, I haven't been able to cause the crash here.
To try to solve this, I've added RESOURCE=OFF to the Config.FPW (to eliminate FoxUser.DBF corruption as a possibility) and checked the code for the RETURN inside WITH-ENDWITH problem (no cases
found). I've now given them a new build with more logging than I had before, and particularly logging some UI actions to the log I'm creating in DL, hoping that'll give me some clues.
I've passed other ideas, like virus scanner issues and driver problems on to the client to try. They did report these configurations for the two users who've already reported crashes with the latest version:
User 1:
Display Adaptor = Intel(R) 82915G/GV/910FL Express Chipset Family
Driver = C:\WINDOWS\system32\DRIVERS\ialmnt5.sys + others
Provider = Intel
File Version = 3.0.0.4020
Copyright = 1999-2004, Intel
Digital Signer = Microsoft Windows Hardware Compatibility
Set to 1080x1024 pixels, 32bit color
Processor = Intel Pentium 4 CPU, 3.20GHz
Memory = 3072 MB
Operating System = Microsoft Windows XP Professional, 5.1.2600
Service Pack 2
User 2:
Graphics = Intel(R) 82845G/GL/GE/PE/GV Graphics Controller
Graphics Driver = C:\WINDOWS\system32\DRIVERS\ialmnt5.sys + others
Provider = Intel
Date = 3/23/2006
Version = 6.14.10.4544
Digital Signer = Not digitally signed
1024 x 768 pixels
Highest (32 bit) color
Intel Pentium 4 CPU, 2.00 GHz, 512 Mb RAM
Microsoft Windows XP Professional, Version 2002
Service Pack 2
Any ideas where to look would be appreciated.
Tamar
Here's the background: I have a VFP EXE (call it UI) that instantiates a VFP out-of-process COM object (call it DL). DL instantiates another VFP COM object (call it OnlineTimer) as well as a C# COM object (call it Comms).
The C# object, Comms, talks to some hardware and passes information between the hardware and DL. When it has information ready for DL, it sets a flag. DL has a timer that interrogates that flag and when there's data available, processes it. Processing the data means parsing the XML format that Comms returns and, generally, poking updates into the object model maintained by UI. Many, many lines of code in DL look something like:
This.oCallingApp.oNetwork.DoSomething()
They mostly drill more layers down into the object model before actually calling a method.
After DL finishes processing one returned item, it makes a call to a UI method that repaints the interface. Each form has a method that can be called to update its display with the latest data.
The C5s are happening in DL. Based on the VFP9Rerr.LOG file that's automatically created, they happen in various places, but it looks like the line that crashes is always one that includes This.oCalling.something. (I don't have enough data yet to be sure that's the case.)
Naturally, I haven't been able to cause the crash here.
To try to solve this, I've added RESOURCE=OFF to the Config.FPW (to eliminate FoxUser.DBF corruption as a possibility) and checked the code for the RETURN inside WITH-ENDWITH problem (no cases
found). I've now given them a new build with more logging than I had before, and particularly logging some UI actions to the log I'm creating in DL, hoping that'll give me some clues.
I've passed other ideas, like virus scanner issues and driver problems on to the client to try. They did report these configurations for the two users who've already reported crashes with the latest version:
User 1:
Display Adaptor = Intel(R) 82915G/GV/910FL Express Chipset Family
Driver = C:\WINDOWS\system32\DRIVERS\ialmnt5.sys + others
Provider = Intel
File Version = 3.0.0.4020
Copyright = 1999-2004, Intel
Digital Signer = Microsoft Windows Hardware Compatibility
Set to 1080x1024 pixels, 32bit color
Processor = Intel Pentium 4 CPU, 3.20GHz
Memory = 3072 MB
Operating System = Microsoft Windows XP Professional, 5.1.2600
Service Pack 2
User 2:
Graphics = Intel(R) 82845G/GL/GE/PE/GV Graphics Controller
Graphics Driver = C:\WINDOWS\system32\DRIVERS\ialmnt5.sys + others
Provider = Intel
Date = 3/23/2006
Version = 6.14.10.4544
Digital Signer = Not digitally signed
1024 x 768 pixels
Highest (32 bit) color
Intel Pentium 4 CPU, 2.00 GHz, 512 Mb RAM
Microsoft Windows XP Professional, Version 2002
Service Pack 2
Any ideas where to look would be appreciated.
Tamar