- From: Florent Georges <fgeorges@fgeorges.org>
- Date: Wed, 25 Feb 2009 20:59:48 +0100
- To: XProc Comments <public-xml-processing-model-comments@w3.org>
Hi, If I am right, for now, if one wants to send the content of a file as an HTTP request's body, she has to read the content as an xs:base64Binary, and copy it into the c:request element to be passed to the p:http-request step. But I wonder if it wouldn't be easier to allow to specify an external source for the body content. XML does not copy very well with binary data, and there is no point to read the data once, then to copy it to an XML document, to finaly serialize it to the HTTP wire. Couldn't we just add an "href" attribute to c:body? For a use case, you can think about the Google Documents List Data API <http://code.google.com/apis/documents/docs/> (if you want to guess what I will talk about at Prague ;-p) using which you can upload a document to the Google Documents service. The request is (could be among others) a multipart POST where one part is an Atom entry XML document describing meta information and one part is the document itself. I can imagine a pipeline that takes a filename as input, computes some meta information and send them with the file content to the server (the document can be for instance a binary MS Excel spreadsheet.) Regards, -- Florent Georges http://www.fgeorges.org/
Received on Wednesday, 25 February 2009 20:00:31 UTC