Just checking the code out in juce_win32_WebBrowserComponent.cpp out of curiosity, and it looks like the LPSAFEARRAY variable “sa” is setup twice… Only the first “sa” is being checked for null to be destroyed, so if there’s post data, the second “sa” will not be destroyed.
Ah, good catch, thanks! Yes, that definitely looks like a typo. Have checked-in a saner version now.