Greetings,
In case this article looks long I have given a
brief summary at the end. The lengthy paragraphs
that are seen below could not be helped. I am
trying to be as brief as possible without
cutting the details.
I downloaded the hello.wml file from the wapit.com
site which is used as a test file for the fakewap
application. I renamed hello.wml to hello.xml and
invoked the same via a Ericsson R380 emulator (the
IP address of the machine where Kannel is installed
was set as the gateway address in the Ericsson emulator)
I got Error message "Browser cant read text/html content"
Next I copied a wap.xml file which was stored in the
$LIBXML_HOME/test folder(I assume this XML file outputs
WAP content) to the location where my Apache http
server looks for requested files. I tried to invoke
wap.xml in the Ericsson emulator, but again the same
result- "Browser cant read text/html content"
I checked the Kannel logs both times, after a
series of Octet strings the ending line was:
"Fetched URL/<xml-file-name> (text/html,charset="")
Refusing to give up I was bent upon making Kannel
extract the WAP content out of any well-formed XML
file.
So I configured the XML engine Cocoon to process
all XML requests sent to Apache. In a sample XML
file lying in the Cocoon folder I replaced all HTML
tags to WML tags and placed it in the DocumentRoot
of Apache. A third invocation of the XML file (.i.e
the sample Cocoon XML file outputting WAP content)
should have made Kannel recognize the output and
succesfully pass it on to Ericsson R380 emulator.
BUT again "Browser cant read text/html content"
Thoroughly frustrated now I wrote a servlet program
which outputs WAP content and invoked the servlet
via Kannel, this time I hit the jackpot Ericsson
R380 displayed the servlet's output. I called
hello.wml (mentioned above) and again Ericsson
is displaying the output.
All these experiments lead me to believe that
whenever Kannel sees a .xml extension in the URL
it fetches the output of the XML file and(even if
it is proper WAP content) passes this output to
the GNOME XML library. This in turn fails the
OUTPUT of the XML files I have been trying to invoke.
Any diagnostic help- as to why Kannel fails to
recognize the content-type of the XML files
and labels the content-type as text/html even
though these XML files output proper WAP
content- would be appreciated.
To help the people here I have drawn a small
diagram which explains the short setup that I
have on my system:
___________ ________________
| ERICSSON| REQUEST FOR WAP CONTENT GENERATING XML FILE | KANNEL |
| R380 |_____________________\\__________________________| ____________ |
| EMULATOR| // | | BEARERBOX| |
|_________| | |__________| |
| |
| + |
| _________ |
| | WAPBOX| |
| |_______| |
|______________|
|
| R
| E
| Q
\|/U
\|/E
| S
| T
|
| P
| A
| S
| S
__________ | E
| APACHE | TO APACHE | D
_____\\____| HTTP |________________//_________________|
| // | WEB | \\
| | SERVER |
XML | |________|
OUTPUT | |
SENT | | XML
BACK /|\ \|/ PROCESSING
TO /|\ \|/ REQUEST REDIRECTED
APACHE | | TO COCOON
| ____|____
| | COCOON|
|____//____| XML |
\\ | ENGINE|
|_______|
I am summarizing the actions once more here:
1) RENAMED hello.wml TO hello.xml FILE. BUT text/html
2) COPIED SAMPLE libxml2 (WAP-GENERATING) XML FILE
TO APACHE DOCUMENTROOT STILL text/html
3) CONFIGURED XML ENGINE COCOON WITH APACHE. IN
SAMPLE COCOON XML FILE REPLACED HTML TAGS
WITH WML TAGS. BUT STILL text/html
4) WROTE A SERVLET PROGRAM, SET CONTENT-TYPE
AS text/vnd.wap.wml AND HURRAH OUTPUT OF
SERVLET SEEN IN ERICSSON R380
5) CONCLUSION: WHENEVER KANNEL SEES .xml
EXTENSION, FETCHES OUTPUT AND PASSES
IT ON TO libxml2. GNOME XML LIBRARY
FAILS TO PROCESS OUTPUT AND LABELS
IT AS text/html
Thanks in advance.
SNODX
