Matthew Graham wrote:
> Alasdair Allan wrote:
>> There are certain VOEvent applications where we're not even using
>> an XML parser because it's too much overhead. There are others
>> where a few hours don't matter, and semantic tools could prove
>> very useful (SNe candidates spring to mind). As ever mandating the
>> use of such to get information out of the packet is something I'm
>> against, so as always YMMV.
>
> So maybe it's time we started seriously thinking about a two-tier
> system based on different serializations of the VOEvent data model.
> The ultrafast layer uses a JSON representation of a VOEvent and TCP/
> V as the transport protocol; the normal layer uses the standard XML
> representation and the suite of protocols that we currently support.
Hmm, interesting. I'm not totally against the idea. We should talk about this at ADASS and the Interop. i think a JSON representation could be useful, although I'm not so sold on taking the XML out of the TCP transport layer.
Al.
PS. Matthew, you still ow me a page or two on the Jabber transport layer for the VOEvent Transport paper for the HTU proceedings. Received on 2007-09-14Z15:24:39