That was the original behavior in versions 1.7 and earlier of HTTPAPI.
However, we had problems with data getting truncated with some servers --
they'd close their sockets and their linger times were set low, so the
data would get lost when they got closed.
With the current method, that theoretically won't happen, because HTTPAPI
initiates the closing instead of the web server.
I wish I could figure out why HTTPAPI doesn't think that the document is
complete? But, of course, when I run it it works perfectly. That makes
it very difficult to troubleshoot.
Unless someone who has the problem could give me access to their iSeries
for me to test it out on? I wonder if I loaded it on www.Netshare400.com
if I could reproduce your problem there... hmm...