Re: files or database for solid pods [Was: Towards Solid Lite]

st 1. 11. 2023 v 18:47 odesílatel Jonas Smedegaard <jonas@jones.dk> napsal:

> Quoting Melvin Carvalho (2023-11-01 18:19:22)
> > Jonas, if we were to agree that a Solid Lite spec should have as few
> MUSTs
> > as possible (without, at this point, making a specific choice).  But that
> > it could have as many SHOULDs or MAYs as people wanted, do you think that
> > would be a reasonable basis on which to proceed?
>
> Certainly.
>
> Personally, what I find sensible is to...
>
> * require RDF/* (i.e. any serialization)
> * recommend JSON-LD, and to only suggest RDF/Turtle
> * recommend content-negotiation
>
> Yes, this means not all user agents can expect to have succesful
> conversations with all servers, even if they all comply with solid-lite.
>
> I see the lack of quaranteed succesful conversation as a tradeoff for a
> lite spec.
>
> Only recommending content-negotiation is to me similar to travelling in
> foreign countries: If you learn some addon languages (equip yourself
> with the capability of content-negotiation) then you raise your chances
> of being able to communicate with those you meet, but there is never a
> guarantee, because the others may not have done the same, and even if
> they have you might end in a "succesful" zero-solution negotiation if
> there was no common match.  Yes, in an ideal world we all have eyes and
> all speak danish, and we would have no need for content-negotiation.
>

Makes sense.  Reminds me a bit of Postel's Law:  be conservative in what
you send, be liberal in what you accept

https://en.wikipedia.org/wiki/Robustness_principle

I've made quite a bit of progress on the pre-draft spec, though I am still
calling it 0.0.0 rather than 0.0.1 so far.

Authz-lite is great.

Cors is simple enough.

The exact syntax of the WebID will be a small amount of work.

The challenge is to find a default authentication method for a webid that
would not be too onerous to program.

We might need to think of something simple, and swappable for testing
purposes.  Or use a strategy from something like passport:

https://www.passportjs.org/

It's quite refreshing trying to figure out which bits to remove, and
keeping the parts that are MUSTs.

I've probably dont most of the easy parts now, and hopefully it will not be
too long before there's an implementable 0.0.1

As I say, picking an Authentication-Lite is the challenge.

>
>
>
>  - Jonas
>
> --
>  * Jonas Smedegaard - idealist & Internet-arkitekt
>  * Tlf.: +45 40843136  Website: http://dr.jones.dk/
>  * Sponsorship: https://ko-fi.com/drjones
>
>  [x] quote me freely  [ ] ask before reusing  [ ] keep private