> should be checked the pointer to r->soc:
> if (r->soc != INVALID_SOCKET) {
>    n = send(r->soc, s, ssz, 0);
> }

Humm, strange - are you sure the httrack headers have been updated ? I have
modified some structure layout quite recently, and this may (or not) be
related.

I don't see obvious faults in the code, and the crash locations (both on
socket i/o and filename handling) are really weird.