Ouch, the ones in the Server, ServiceRequestHandler and Proxy for XMLRPC are definitely just for debugging and should not have been left turned on. I have just commented those ones out and checked the changes back in. That leaves just two active uses of f_p_c:

1. in the serviceRequestHandler for jsonrpc
2. in the serviceRequestHandler for soap

we write these into the file system for two reasons - 1 to save regenerating them and 2 for the soap server we need some wsdl to pass to the php_soap extension, and as a file is the simplest way to do it. Actually our logic is suspect in these cases as we don't regenerate when the file changes. But anyway, for the moment we need to keep those two. 

And, WRT the logger we should not be creating a temp directory when we don't need to. That was also a bit poor. I will fix that next.

I've just moved the logic that decides the location of the log by creating a temporary file into SCA_Logger::startlog(), so it doesn't happen if the logging is off. 

I am just in the process of making a new release, so these changes will get into that. I was going to look at your other defect first to see if I can slip that in quickly. 

Thanks for telling us about these problems. If you have other ideas and / or requests do let us know: if you want developer access and want to fix the code too, we can arrange that and you'd be welcome.

This one was fixed a few weks ago, so closing.