Hi

We are getting a StackOverflowException when using the PdfExporter.Export method.  It's only happening on some diagrams and I'm guessing it's the size of the diagram that's causing the issue?  Is that something that you have seen before?

The diagram itself has 164 nodes and 163 links, but is a very flat pyramid, so its very wide and only 3 rows of nodes high.  The export runs for around 17 minutes and then crashes with this error:  "An unhandled exception of type 'System.StackOverflowException' occurred in mscorlib.dll".  I'm not quite sure why it would be taking 17 minutes either?

We are currently running version 5.7.0.25317 and ideally I wouldn't do a version change right now but I can do if there is a chance that will fix the issue?

Thanks, Tom

Hi Stoyan,

Sorry for the delay, file attached.

Apparently this happens because of some strange symbol in node 10285. It causes a stack-overflow in the auto-detected Encoding when PdfExporter calls Encoding.GetBytes(). The symbol appears at the apostrophe position on the second text line - in the "organization?s" string. If you take a look with a hex editor, the Unicode sequence there is EF BF BD.

If you set PdfExporter.DefaultEncoding to Unicode, the exception won't happen, but Adobe's PDF Reader still won't render the node's text, so you better replace the symbol with the normal apostrophe.

I hope that helps,
Stoyan

I've found some further information; EF BF BD is the Unicode replacement character, representing characters "whose value is unknown or unrepresentable" in Unicode:
http://www.fileformat.info/info/unicode/char/fffd/index.htm

I'm not sure how it could show up in the diagram texts though... I suppose it might appear after copy/paste betweeen documents in different languages, and then perhaps the replacement symbol itself was copied to the diagram. Otherwise the nodes' texts should always be in Unicode as they are .NET strings, so copying/pasting ordinary text from any language to the diagram shouldn't cause problems.

Was that data imported from Excel automatically using some tool? I suppose that symbol might appear if text bytes were type-cast directly to (char), instead of using conversion methods of the Encoding class corresponding to whatever encoding is used in the Excel document.