- From: Kev Kirkland <kev@dataunity.org>
- Date: Fri, 11 Nov 2016 11:40:23 +0000
- To: Ruben Verborgh <Ruben.Verborgh@ugent.be>
- Cc: Hydra <public-hydra@w3.org>, Graham Conzett <conzett@gmail.com>, Tomasz Pluskiewicz <tomasz@t-code.pl>
- Message-ID: <CAPNZP6+dU0A3ioexbh6CExr3usmy36wecWu4pPUrr3YTUYV04Q@mail.gmail.com>
Hi all, Hydra is great and definitely going in the right direction, but I agree that a review of the overall architecture would be beneficial too. There's been some developments in other groups since we started which might help. One of the areas I've had problems is with constraining choices in requests - we might want to look at replacing/augmenting Hydra:Class with the work the Shapes Working Group are doing (e.g. https://www.w3.org/TR/ shacl/). I've also had a problem finding ways to navigate through a sequence of steps in an API (e.g. when you need to do multiple form posts in a sequence). Interesting to see that Mike Amundson has been compiling some patterns for hypermedia - I think we could benefit from the navigation patterns: https://www.oreilly.com/learning/12-patterns-for-hypermedia-service-architecture Cheers, Kev On 11 November 2016 at 10:03, Ruben Verborgh <Ruben.Verborgh@ugent.be> wrote: > Dear all, > > Yes, we need to do something to bring the ideas in Hydra back to life. > > At the same time, I think we need a grand architectural vision for all of > this. > Only bottom-up, issue-driven improvement will not get us there, I'm afraid. > > I think we need to make a clean start and do things the right way: > start from the scenarios we which to enable, > define concrete use cases and examples of how Hydra should be used. > > Then, we should work our way top-down, finding the larger pieces we need > (two distinct pieces being API description and hypermedia controls, IMHO) > and than gradually going down, perhaps with different subgroups > working on different pieces of varying granularity. > > I believe the current Hydra specification was great as a working document, > and has spawned several ideas and discussions, > but I'm afraid we'd become stuck in such discussions > if we don't get the grand design right upfront. > We now have more experience to get this right > than we had when this was all starting, > so I propose we all take the opportunity to do so. > > Best, > > Ruben > -- www.dataunity.org twitter: @data_unity
Received on Friday, 11 November 2016 11:40:59 UTC