Re: urn:solid: for prototyping predicates

Dear all,

Am 26.05.25 um 09:56 schrieb Melvin Carvalho:

> po 26. 5. 2025 v 9:33 odesílatel Christoph Braun <braun3@fzi.de 
> <mailto:braun3@fzi.de>> napsal:
>     On 25/05/2025 08:54, Melvin Carvalho wrote:
>>     čt 22. 5. 2025 v 15:32 odesílatel Christoph Braun <braun3@fzi.de
>>     <mailto:braun3@fzi.de>> napsal:
>>
>>         I would like to propose dogfooding our own technology.
>>
>>         An app developer wants to build an app:
>>         - they search the Solid Catalog (or similar to other vocab
>>         repos [1] ) for suitable vocabularies
>>         - if they do not find a suitable vocabulary, they create a new
>>         vocabulary
>>         - they use their Solid Pod to host the new vocabulary
>>         - they Link it in/from the Solid Catalog for discovery
>>         - so other developers can search the Solid Catalog, discover
>>         the vocabulary and re-use it in their app
>>
>>     Just a small note on the idea of dogfooding vocab publishing via
>>     personal Pods. It’s a lovely ideal, but maybe a bit too heavy for
>>     many devs.
> 
>     It is lovely indeed.
>     In particular, because every developer who wants to build a Solid
>     app can go to solidcommunity.net <http://solidcommunity.net> and
>     create a Pod there.
>     Getting a (hosted) Pod is not a problem.
>     If afraid of tying the vocabulary to a Pod Provider's domain, use
>     w3id or purl to re-direct whereever the vocabulary is currently hosted.
> 
>     For development, just using a test account on solidcommunity.net
>     <http://solidcommunity.net> is much more convenient than hosting you
>     own local Pod server instance.
>      From my project experience (see presentation at Solid World [1]), a
>     developer of a Solid app creates a test account for testing their
>     app at some point anyway instead of going through the hassel of
>     learning how to setup a local Pod.
> 
> 
> I would caution against this approach to hosting vocabularies.
> 
> - That site is not a CDN, it's a volunteer-run VPS. Heavy traffic can 
> hit its acceptable-use limits and add latency for many users.
> 
> - Using it for lots of vocabs becomes DDoS vector, which can impact 
> regular users

How often do vocabulary terms get dereferenced? Maybe somebody from 
w3.org can shed some light on how much of a burden it is to serve [A]. 
At least, schema information is pretty static [0], so there is ample 
opportunity to use the caching functionalities of the Web.

> - If it runsout of funding, then there are no “Cool URIs.”

You can also host your vocabulary on GitHub or GitLab pages, domains 
with some commercial interest to stick around for a while (and you can 
even bring your own domain such that you can move elsewhere if they turn 
rogue). I read that meanwhile they supply the right Content-Type header 
for Turtle files [3].

Back in the day when they did not have Content-Type text/turtle for .ttl 
files, the closest I got for standards conformance was RDFa in HTML [1]. 
This requires devs to write a trivial YAML file and a tiny bit of HTML, 
and it looks more shiny to the eye non-initiated to the beauty of Turtle.

So my +1 for dogfooding and self-hosting vocabularies, there are many 
ways to achieve it, even for the non-expert.

Regards,

Tobias

[A] https://www.w3.org/1999/02/22-rdf-syntax-ns
[0] Käfer et al: "Observing Linked Data Dynamics!" Proc. 10th ESWC, 2013 
https://link.springer.com/chapter/10.1007/978-3-642-38288-8_15
[1] https://kaefer3000.github.io/coinflip-servlet/vocab
[2] https://github.com/kaefer3000/coinflip-servlet see _config.yml and 
vocab.html
[3] https://github.com/mfhepp/test_mime_types

Received on Monday, 26 May 2025 08:31:38 UTC