W3C home > Mailing lists > Public > www-annotation@w3.org > January to June 2002

annotea HTTP library changes

From: Eric Prud'hommeaux <eric@w3.org>
Date: Sun, 24 Mar 2002 17:20:50 -0500
To: www-annotation@w3.org, Art Barstow <art.barstow@nokia.com>
Message-ID: <20020324172007.A26824@w3.org>
This text describes some annotea protocl issues and the affects on
deployed clients. Users of the annotea servlet [1] may wish to read a
section at the bottom: Deprecated Protocol.

The library that annotea uses to get HTTP protocol data (ie, the query
strings, POST data, etc.) was getting clouded in heuristics. In the
process of overhauling, I uncovered some annotea protocol issues:

The are two encodings for POSTed annotations, application/xml and
url-encoded. The url-encoded format expects a w3c_annotate parameter
with a value of the RDF/XML-encoded annotation. In addition, it
permits the passing of additional parameters, for instance
replace_source and rdfType for replacing annotations.

application/xml data is defined to be the w3c_annotate parameter.
Additional parameters may be encoded as CGI parameters appended to
the annotate script's URL.

There is a lack of parallelism in this situation. url-encoded data
requires a parameter name for the submitted RDF while the application/
xml data has a defined parameter assignment for the payload. One way
to view this is, application/xml requires auxilliary communication of
additional parameters while application/x-www-form-urlencoded imposes
an additional layer of encoding, capable of communicating an arbitary
number of parameters.

It would be possible to defined the protocol such that the payload of
url-encoded messages would be assumed to be the w3c_annotate parameter
and that all auxiliary parameters be passed in the POST URL (as is
done with application/xml data). I believe this would be ill-advised
as rfc1866 states that url-encoded data is {parameter: value} pairs.

from rfc1866 [2]:
8.2.1. The form-urlencoded Media Type

   The default encoding for all forms is `application/x-www-form-
   urlencoded'. A form data set is represented in this media type as

        1. The form field names and values are escaped:

rfc1866 does not address the issue of mixing POST and GET data (ala
the application/xml data POSTed to create annotations) as it is
primarily an HTML language specification. This mixing is not possible
with HTML forms.

If we were to decide that we needed a way to encode multiple parameters
in the application/xml data, we could use something like SOAP [3] or
parseType=Literal [4] or reification or maybe just use magic.

Deprecated Protocol:
The current implementation of the annotea servlet sends a url-encoded
payload with no parameter name. This special case is handled by the
annotea script:
    # @@@ temporary hack to deal with clients that POST urlencoded data without CGI parms
    if ($ENV{'REQUEST_METHOD'} eq 'POST' && 
	$ENV{CONTENT_TYPE} eq 'application/x-www-form-urlencoded' && 
	$self->{READ}->getPOST() =~ m/^<\?xml/) {
	$ENV{CONTENT_TYPE} = 'application/xml';
	$self->{RDF_INPUT} = &CGI::unescape($self->{READ}->getPOST());
	# this hack requires the parms be ignore (as they are garbage).
I wish to remove this code once the clients using that protocol are
updated. Other than the annotea servlet, what other apps use un-
parametered names in url-encoded POST bodies?

[1] http://www.w3.org/2001/Annotea/Bookmarklet/Annotea-JavaScript
[2] http://www.ietf.org/rfc/rfc1866.txt
[3] http://www.w3.org/TR/soap12-part0/
[4] http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/#parseResource

Feel free to forward this message to any list for any purpose other than
email address distribution.
Received on Sunday, 24 March 2002 17:20:52 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:16:55 UTC