Document Conversion System Woes

[Bitwise]

I am working on an online document conversion system using OpenOffice as the engine. Most of the documents we receive are MS Word, and the majority are converted perfectly. A small amount of them cause the OpenOffice instance running to take up all system resources, hang and not respond.

I have coded both a macro and C++/UNO program to do the task, and while the C++ program catches more of the errors, there are still a few key documents that cause it problems. No exceptions are raised. When i view the graphical output when trying to load these documents, the program window is just blank.

The program hangs when I call loadComponentFromURL in both the Basic macro and the C++ app. I am having a difficult time finding a complete set of MediaDescriptors, so I am not sure if there is a remedy with those settings.

I don't need the system to be able to convert every single document. What I need is for the program to give up gracefully at the right time.

[hol.sten]

[SergeM]

[Bitwise]

Yes, i have a handful of test documents that have choked the system, and i test both solutions with those.

I have also noticed, and perhaps this is a resource issue, that i cannot convert two documents at once.

My current strategy is set up a conversion queue. I had the same idea as hol.sten, to set a watch thread, and time out if needs be. I have implemented this in a bash script, running the conversion as a background process. It dies rather gracefully.

I did some research on setting up a M$ server with some of the readily available word doc conversion software, and discovered that they all depend on a printer driver to convert to pdf, so none of them can handle more than one document at a time, either. Not to mention the small fortune people want for that stuff.

I'll stick with the open source.
_________________
...perhaps if i wave this prop and technobabble at it for a moment, i can formulate a plot device...