- From: <sdo@novell.com>
- Date: Thu, 21 Mar 1996 16:50 EST
- To: MACRIDES@sci.wfbr.edu (Foteos Macrides)
- Cc: www-html@w3.org
>Date: Thu, 21 Mar 1996 10:29:20 -0500 (EST) >From: MACRIDES@SCI.WFBR.EDU (Foteos Macrides) >Message-ID: <01I2LICAGGW2007392@SCI.WFBR.EDU> > > > Bearing in mind that the INSERT draft is in the process of >being revised, it already enables insertion of static or dynamic >HTML fragments, with no need to modify servers if that's what's >providing the fragments, and only minor mods needed for the client >software. You can insert static fragments via: > ><INSERT > type="text/chtml" > data="http:/host[:port/path/foo.chtml"> >Altenative markup, or zilch, for INSERT-blind clients. ></INSERT> > >where .chtml has been mapped to text/chtml in your server's configuration >file. > > For dynamic inserts, you can use: > ><INSERT > type="text/chtml" > data="http:/host[:port/script_path[/path_info][?pseudo_query]"> >Altenative markup, or zilch, for INSERT-blind clients. ></INSERT> > >for invoking a CGI script, with PATH_INFO and QUERY_STRING >environment variables for instructing the script on what >fragment to create. Unfortunately, I am trying to define the substitution language for the output of a CGI program that has already been called by another FORM or even INSERT tag, as you have described above. I would like to input a request on a FORM, which submits a POST or GET query to a Web "service." Although I have not said that the service protocol must be CGI, I have said that the data format is convenient for many data transfer applications (FIELD=VALUE pairs). I don't necessarily want the output to appear on the input form. I also do not want to be tied to HTML. There's no reason why my language couldn't be used as a pre-gif or pre-txt, rather than just pre-html. To refine my proposal a bit, let's assume the native Content-Type, but a Content-Encoding of "field-substitute". However, I doubt whether the content encoding can be passed into the HTTP_ACCEPT header, such that a script or server could decide whether or not to expand the output itself. > > You also can use other standard types, not necessarily a >"made up" one for CSIs, but point via data= to a script which >handles what's returned dynamically. For "text/chtml", the >client needs some minor initialization and configuration tweaks >for recognizing and handling that as an HTML fragment insert (and >confidence that the fragment won't create bad HTML within the >surrounding markup). Most importantly, the client also needs tweaks >not to include the fragment in any file cache it creates for the >overall document, and that's simple to "engineer" for a distinct >type (e.g., "text/chtml"), but some clear attribute to indicate >that for standard types is needed. My goal is to only send back the stuff to be included (the includ-ee) rather than the document that will later include some other stuff (the includ-er). Some external mapping specifies (perhaps based on data or service name content) the includ-er document, in which some, all, or none of the includ-ee fields are substituted. It's an include-me protocol, rather than an include-this protocol. > > Fote > >========================================================================= > Foteos Macrides Worcester Foundation for Biomedical Research > MACRIDES@SCI.WFBR.EDU 222 Maple Avenue, Shrewsbury, MA 01545 >========================================================================= > Scott Orshan BEA Systems (Formerly Novell) TUXEDO Engineering Group +1-201-443-5063 sdo@novell.com
Received on Thursday, 21 March 1996 17:16:37 UTC