[Laszlo-dev] An LPS server data proxy protocol for raw data, not just XML?

P T Withington ptw at pobox.com
Wed Jul 30 04:38:03 PDT 2008


I think you are saying that currently DATA (in your example below)  
must be well-formed XML and you want a way to send random data.  So,  
don't you just need an encode/decode step that encodes random data  
into a narrower symbol set?  E.g., (HTML) escape would suffice for  
ascii data, or BINHEX for binary.  Use different tags to say the  
content needs to be decoded first (and what the decoder is).

<resultset><escaped>ESCAPED</escaped><headers>HEADERS</headers></ 
resultset>

<resultset><binhexed>BINHEX</binhexed><headers>HEADERS</headers></ 
resultset>


On 2008-07-29, at 16:52EDT, Henry Minsky wrote:

> The protocol we use for proxied data requests is  really only suitable
> for requesting
> well formed XML data. With the current protocol, the data comes back
> wrapped inside an XML document,
> which contains optional header data, and is assumed to be well  
> formed XML.
>
> e.g.,
> <resulset><body>DATA</body><headers>HEADERS</headers></resultset>
>
>
> As we want to support non XML protocols, like JSON, or raw text,
> which may need to be proxied
> through a server for security or other reasons, it would be nice to
> make the server proxy protocol
> support fetching arbitrary string content.
>
> My first thought is to add an extra flag to the request query args,
> which is something like "raw=true",
> and then the server would just proxy the request and send back the
> response verbatim. This
> would be simple, but would not allow any way to get the proxied
> response headers.
>
> I guess another idea would be to proxy the HTTP response verbatim,
> with the headers separated from the
> body by two blank lines. The client would need to parse as far as
> that, and could then take the rest of the
> response using substring(...).
>
> Any thoughts?  Am I looking for a solution without a problem?
>
>
>
>
> -- 
> Henry Minsky
> Software Architect
> hminsky at laszlosystems.com



More information about the Laszlo-dev mailing list