On Sun, 2 Nov 2008, Tim Roberts wrote:
> scott212 <[EMAIL PROTECTED]> wrote:
> >
> >I'm responding with xml to a web request from google checkout but I
> >think I'm in a catch-22. To get my webserver (apache) to respond I
> >need a header and then a blank line before my xml begins, or else it
> >throws the 500 error. If have the blank line before the xml, the
> >service I'm responding to cannot parse it. If it's a response, can I
> >still use urllib or something like that? Is that my answer?
>
> You must be misunderstanding something. The HTTP protocol REQUIRES that
> the headers and the content be separated by a blank line.
>
> Perhaps you should try to use a debug proxy to intercept the exact text of
> the response, just to make sure you're sending what you think you are
> sending.
> --
> Tim Roberts, [EMAIL PROTECTED]
> Providenza & Boekelheide, Inc.
>
I agree with Tim, check the output. I had the same issue while serving up
GIF images generated on the fly. I was sending two blank lines before the
image data since an extra newline was being automatically added by the
"print" statement.
# Print the header information
print "Content-type: image/gif\n\n"
The above fails since it sends: "header,newline,newline,newline" with the
last newline appended by the print statement. A solution would have been
to simple remove one of the my coded newline characters and let the print
statement add the second one. However, I switched to "write" statements so
I could _see_ the newline characters in my code.
import sys
# Write out the header information
sys.stdout.write("Content-type: image/gif\n\n")
The above worked because "write" does not add an automatic newline
character to the output.
Also, make sure your content-type is correct. The same content would be
handled differently by a browser depending on how it is identified.
"Content-type: text/html" - Sent through the HTML parser then displayed.
"Content-type: text/plain" - Displayed without formatting.
What is the application you're communicating with looking for?
Hope that helps,
Christopher
--
http://mail.python.org/mailman/listinfo/python-list
