Re: Low power devices (was: webid serializations consensus 2023)

st 5. 7. 2023 v 16:32 odesílatel Jacopo Scazzosi <jacopo@scazzosi.com>
napsal:

>
> > Firstly let me point out that linked data is all about fetching data
> from the web via hyperlinks
>
> In principle, yes. In practice, I find that LD approaches, techniques,
> methodologies can bring enormous value in intranet contexts, too. LD can
> prosper on closed networks and I think corporate adoption of WebID would go
> a long way in making it a standard.
>

JSON-LD can be in different forms or "profiles", when needed.  These
present different parsing shapes for specific use cases.


>
> > Most specifications do not point out that the context can be inlined.
> This is of course superior in many ways because it removes one moving
> part.  So in order to facilitate use cases that are self contained, we
> could try and reach consensus on some text that would explain that the
> context can be inlined and how to do that.
>
> Agreed, but if WebID were to mandate (as in MUST) support for JSON-LD,
> context inlining would often become an additional constraint required for
> adoption of the spec while being external to spec itself. An additional
> constraint that arises so early in the process gives me pause, although it
> might very well be ok in the end.
>

That's why I pointed out that inlining is possible in a separate thread.

I was just pointing out it's another tool that can work in specific use
cases.


>
> However, this being a thread about low-power devices, what worries me the
> most when it comes to JSON-LD is the sheer performance and complexity
> difference between JSON-LD libraries and those of other formats. Granted,
> this also comes down to the level of optimization of any specific library
> but, IMHO, the overall state of the ecosystem must be taken into account
> when choosing whether to mandate a format or not.
>

Profiles, as above


>
> Implementing a performant Turtle parser / serializer is something I am
> confident I could do on my own if I ever needed to. On the flip side, I do
> not think I would be able to implement a compliant JSON-LD parser /
> serializer, let alone a performant one. I could likely implement a parser
> for a subset of JSON-LD but then we get back to my first point about
> additional constraints.
>

You would use a library, or make use case specific examples.  If WebID
takes TallTed's definition, or something similar, then there can be dozens
of WebID use cases, that dont have to be listed in the main spec.


>
> All this said… I am aware that my use corporate + iot case is a niche one,
> at least for the foreseeable future. I don’t expect this group to embrace
> it, although I appreciate the chance to argue for it.


Im saying that it's been embraced, used, and widely adopted already.
Schema.org is used in billions of pages.  It's not niche, it's so
matinstream, that it's the de-facto semantic web, right now

But it IS JSON-LD.  And that is missing from the webid spec.  And if we
dont add it in, it might not make it to the final draft.