Sean just found a bug in the OSCAR file transfer timeout code today. It turns out that the libFaim code I was using to identify the file transfer connection as successful or not is not reliable for my purposes. Instead, I have now added a variable to the oft_info structure to track whether or not the connection was successful. That pretty well squished the bug. I realized as I was reading Sean's email the mistake I had made in my testing: I tested on a few different computers and different clients and even different files... but they were all on my home subnet. Even though I used file that were several megabytes in size, they still transfered at around 3MB/s, so the error didn't have a chance to show its ugly head. This is one mistake that I will not repeat; I am now using a 1.9GB file to test all code before doing a CVS commit.

Also, since my last post, I have successfully received a file via a "stage 2" AOL proxy. This just leaves a stage 2 file send left on my grocery list of proxied transfer scenarios that must be covered. After that, I'll just have to tidy up the code a bit, add some sort of option for the user to define whether or not an AOL proxy should be used (hopefully), and then write up documentation.