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

út 31. 10. 2023 v 22:46 odesílatel Nicolas Chauvat <
nicolas.chauvat@logilab.fr> napsal:

> Le Tue, Oct 31, 2023 at 04:40:31PM -0400, Kingsley Idehen a écrit :
>
> > Solid is supposed to be about choice on both the client and sever sides
> with
> > regards to read-write operations. It will always lose its way when
> > specificity (on either side) creeps in i.e., whenever the abstraction
> > becomes leaky.
>
> +1
>
> When I do an HTTP GET on some apache server configured to serve a file
> system, I get HTML.
>

An apache server would give an octet stream unless you request an .html
file or it was configured to return html

We could do that


>
> When I do an HTTP GET on some search engine, I get HTML built on the
> fly to show the results from a database query.
>

That's also fine, from the client point of view, it does not need to know
the implementation


>
> My browser wants HTML and will not care if that HTML existed as a file
> before my GET or was built right after my GET arrived at the server.
>

Correct, from the browser perspective


>
> Do you think we could have the same with the abstractions of Solid ?
>

Absolutely yes.

But there also needs to be instructions for a server, and how to build one,
servicing these requests.

Ability to quickly create a Solid Lite server MVP will have a close
relationship with the initial feature subset that is selected as mandatory.

However, that's just the HTTP part.  The file system is also part of the
design architecture of solid, to take advantage of things such as the file
system API, where you can write to your local disk, rather than use HTTP
PUT.


>
> --
> Nicolas Chauvat
>
> logilab.fr - services en informatique scientifique et gestion de
> connaissances
>
>