John Wilson wrote: > Measuring the time to parse is also measuring the time to read the > document. I would guess we are seeing some sort of network problem here.
Quite true. In which case ethereal or tcpdump might prove that there is such a problem. Jochen
