- From: Ruben Verborgh <ruben.verborgh@ugent.be>
- Date: Wed, 24 Jun 2015 16:52:41 +0200
- To: Luca Matteis <lmatteis@gmail.com>
- Cc: public-linked-data-fragments@w3.org
Hi Luca, > Have you thought about a triple pattern fragments interface that is > exposed using WebSockets rather than HTTP? On the server it would be > the same amount of work (just matching triple patterns), while on the > client, it could be less resourceful since you'd instantiate a single > request (rather than one for each pattern). I hadn't thought about WebSockets yet, but I was thinking about HTTP/2. I assume that there would be large benefits there as well. > I'm not sure however if ws:// links are considered linked data, since > you wouldn't have a ws:// link for each resource. Addressability would seem to be the main concern, but also cacheability would be important here. HTTP/2 would be able to provide both, perhaps with high performance. Best, Ruben
Received on Wednesday, 24 June 2015 14:53:11 UTC