Re: Hydra Status

November 15 2016 9:55 PM, "Markus Lanthaler" <markus.lanthaler@gmx.net> wrote: 
> Very interesting discussion folks. To briefly summarize it, the main points that came up in the
> discussions in the last days was a demand for more examples, tutorials, and comparisons to related
> efforts and a desire to move discussions from the mailing list to GitHub. Last but not least, a
> reexamination of the overall architecture came up.
> 
> I have a slightly different perspective on this... and also a different focus. IMO we are not at
> the stage yet were we should be concerned about adoption. I thus don't see work on examples,
> tutorials, comparisons etc. as a high priority yet. They are certainly useful in evaluating designs
> etc. but they require lots of work and hour resources are better spent elsewhere at the moment IMO.
> Keep in mind that till the spec is stable the cost for creating up to date examples is much much
> higher. 

I see where you are coming from. On the other hand there are problems like what I'm facing. I'm trying to slowly implement my Hydra library and battle test it by creating a new console and reusable components. I was about to start implementing collection filtering (I already have collections based on the _view_ draft) and I realized that the spec is so incomplete. Not to mention a multitude of competing and incompatible proposals. 

Bottom line is that until we actually start eating our own dog food, we will not be able to come up with good design. And too many discussions here were very much theoretical and going into minuscule details. I think that it'll be easier to actually deliver when we focus on actual use cases based on real code. Even if it's demos for demos and experiments. 

> 
> What I think we are sorely missing at the moment, is a group of *dedicated and committed* core
> contributors that push the development of Hydra forward. And by this I don't only mean taking the
> time to deeply think about designs etc. What's probably more important is that (at least) core
> contributors get familiar with the underlying technologies. Hydra is built on JSON-LD which in turn
> is based on RDF's data model. It is supposed to work well with HTTP and other web technologies. I
> have been trying to be as welcoming as possible to onboard people not familiar with those
> technologies. I have explained very basic stuff over and over again. Maybe that was
> counterproductive in hindsight as it sidetracked lots of discussions. Looking back, I also realize
> that we had lots of great discussions but only very few concrete proposals have been made.
> Unfortunately I have no solution to all of this so I would like to hear your thoughts on these
> topics.

Dedication is one thing. There have to be a number of people that will actively respond to issues and questions.

Regarding Hydra being entrenched in JSON-LD nad RDF, the only real way is to have examples. Examples of how and why Hydra solves people's problems. The current spec and other resource don't make it clear enough IMO. 

> 
> To address the other two items I mentioned at the beginning of this mail... I'm open to move the
> discussions to GitHub as an experiment, but we need to somehow export and properly archive the
> discussions we have there. This is very important from a standardization point of view. Likely
> other WG/CG have already built some solutions for that in the meantime (back in the JSON-LD days
> there was nothing readily available). 

As a proponent of moving to GitHub I'd like to restate and clarify my stand. I'm not proposing to abandon the mailing list. By no means. However it's been my impression from day one, that using a mailing list limits visibility. It's harder to follow discussions, embed code snippets and search past topics. GitHub is made precisely to discuss technical issues and track them wrt to the actual spec. Linking between issues would also be a plus. Currently we are completely out of sync between the mailing list, GitHub issues and what has actually been included in the spec.

The biggest pro of the mailing list is that all registered members always receive those emails. It is of course possible to watch a repository but it's not the default. I probably would like to keep using it for "general" discussions like this one.

> We can also revisit Hydra's overall architecture but I'm a
> bit skeptical about such an effort if we can't reach consensus on something as fundamental as
> collections.
> 

I don't think that's necessary. We should focus on missing features but keep them simple enough and extensible. We are definitely on good track but progress has stalled somewhat and it could send the wrong message to wider community. Not only the W3C bubble but developers who come across Hydra/JSON-LD/RDF for the first time.

Thanks,
Tom

Received on Thursday, 17 November 2016 09:24:17 UTC