- From: Markus Lanthaler <markus.lanthaler@gmx.net>
- Date: Thu, 23 Jan 2014 16:48:11 +0100
- To: <public-hydra@w3.org>
- Cc: "'James Langley'" <jdhlangley@gmail.com>
On Wednesday, January 22, 2014 8:58 PM, James Langley wrote: > I'm looking into the possibilities for implementing a RESTful API at > the moment. We've been thinking about what data format / media type we > would use. The two front-running candidates for us are Siren and > Hydra. Neither seem to have great tooling support or plentiful > brilliant examples to learn from at the moment. What kind of tooling would you need/like to have? What programming language are you using to implement your API/clients? > I'm wondering whether the fact that JSON-LD has become a standard will > make it more likely that tools will be developed for it, which could > help with getting tools in place for Hydra; or whether the simpler > format of Siren and existing libraries (such as Siren4J) means we will > see more of a community forming more quickly around Siren. I don't know but I think that being a W3C standard and having already been adopted by Google, BBC, Globo.com and the likes JSON-LD and thus indirectly Hydra has some advantages here. I agree that Siren is slightly simpler but that results, in my (biased) opinion, in degraded extensibility and reusability. You basically end up creating snowflake APIs. All of them are similar but don't share concepts such as classes or fields (in Siren-speak). I know Kevin is on this list and I'm sure he'll chime in if I'm wrong. Hydra, on the hand, hugely bets on reusability and linked data principles. Often you won't need to define your custom data model. Instead, you would just reuse something like Schema.org. All that's left for you, is to describe the behavioral model (as I like to call it). This has the advantage that the contract is owned by a third party and thus reduces the risk to expose implementation details or to tightly couple clients to the server. They are coupled against a shared vocabulary instead. I gave a talk about this last year, both the video [1] and the slides [2] are online. For a more abstract discussion, you may be interested in the paper at [3] and the corresponding slides [4]. > There are of course other factors in the decision besides tooling > support. Definitely. Cheers, Markus [1] http://bit.ly/sl-portland2013-video [2] http://m.lanthi.com/sl-portland2013 [3] http://m.lanthi.com/wsrest2013-paper [4] http://m.lanthi.com/wsrest2013-preso -- Markus Lanthaler @markuslanthaler
Received on Thursday, 23 January 2014 15:48:43 UTC