>>Now, I'm afraid there may be another problem (it existed before the fix):
>>if _both_ URLs don't exist (in Dmitriy test code) and FAILONERROR is on,
>>the CPU goes to 11 :-)
>
>Not for me. (and FAILONERROR makes no difference for FTP). Can you tell us
>some more details about the test program and how it differs from test case
>533?

OK, actually I'm interested in HTTP. Since I'm lazy I slightly modified
Dmitriy's initial program, see attached. For me it hangs at 100% CPU.