Re: Post vs Post_raw

[Scott Klement <sk@xxxxxxxxxxxxxxxx>]

Sender: Scott Klement <sk@xxxxxxxxxxxxxxxx>


> 1) Since the post programs are our first try connecting to the parent 
> company's web server and we got a timeout error when the web server was 
> hung, can we rule out connectivity problems?

If you're getting a '500' error from the server, then this isn't a 
connectivity problem.  If the server is able to send you an error, you're 
obviously communicating with it.


> 1) Do you know why the http_url_post program hangs while the 
> http_url_post_raw program did not?  It seems that the post (no raw) is 
> expecting a document that the web server isn't giving it?

Heh...  Shouldn't this be question #2?  (You started two questions with 
"1)")

Actually, post() and post_raw() are the same routine under-the-covers. 
The only difference is that post() calls IBM's write() API (via procedure 
pointer), whereas post_raw() calls a routine of your devising via that 
same procedure pointer.

Since the same piece of code is communicating with the server in both 
cases, I don't see how there could be a difference on your end.  The 
difference is on the server-side.


> 2) For this project, we don't need to process the data received back 
> from the server, so would you suggest we use the http_url_post or the 
> http_url_post_raw function?

I'd suggest the _raw() routine in that case, since you can choose to 
completely discard the data instead of saving it somewhere.


> 3) Why does the HTTP API utility check for the length of the data 
> returned from the web server?

To make sure that it received everything.  Otherwise, if a network error 
caused you to receive part of the data, but not all of it, the program 
wouldn't be able to detect that something went wrong.


> Is this HTTP 'under the covers' best practice?  My first thought, when 
> I saw this code, was "how does the utility know that it will get any 
> data back from the server, let alone that it will be longer in length?"

The server sends a keyword that says "Content-Length: 1234" so that HTTP 
clients will know that they should receive 1234 bytes in response.

> If successful, the java servlet returns the message (data) back (we see 
> this in the GUI).  Is the "Not all data written!" error caused by the 
> servlet sending because the RPG program is sending 'msg=iSeriesTest' and 
> the servlet is sending something back something smaller?

My guess is that you're not encoding the data that you're sending to the 
servlet in the manner that it expects you to.  This is causing the servlet 
to crash or get "hung up" so that it stops receiving data from you.

Then, HTTPAPI complains that the server won't receive all of the data that 
you're trying to send.


> 4) What suggestions do you have to resolve this issue? I was thinking 
> about commenting out the length checking logic from the HTTPAPIR4 member 
> but don't know if that would cause other HTTP errors, so wanted to check 
> with you first.

There are many different ways to encode the parameters that you're trying 
to send to the server.  Just because my server expects a string like 
parm1=value1&parm2=value2&parm3=value3  doesn't mean that your servlet 
does. It may be expecting a SOAP document, or some other XML document, or 
it may be expecting something else entirely.

I'd run a packet sniffer (www.ethereal.com has a good, free one) to find 
out exactly what your test program sends to the server, then make sure 
that HTTPAPI sends the same thing.

-----------------------------------------------------------------------
This is the FTPAPI mailing list.  To unsubsribe from the list send mail
to majordomo@xxxxxxxxxxxxx with the body: unsubscribe ftpapi mymailaddr
-----------------------------------------------------------------------