- From: Sarven Capadisli <info@csarven.ca>
- Date: Mon, 7 May 2018 11:32:34 -0400
- To: semantic-web@w3.org
On 2018-05-07 09:58, ajs6f wrote: > Trellis LDP provides similar equipment, including a standalone server: > > https://github.com/trellis-ldp/trellis/tree/master/platform/linux > > which also creates rpm and deb artifacts, a WAR deployment: > > https://github.com/trellis-ldp/trellis/tree/master/platform/webapp > > and Docker gear: https://github.com/trellis-ldp/trellis-deployment > > I've actually been quite impressed over the last few years at the way that many semweb and LOD projects have taken up devops ideas and made their deployment much easier. Oh, trellis is further ahead in about every aspect than mayktso, and probably many other LD(P)-and-friends implementations. "Similar equipment" (and quality) is an understatement :) For anything super serious / production level, I would not recommend mayktso; it was originally intended to get LDN (test suite) going and to test things with dokieli. The Fedora implementations are looking great,.. and with the spec in the works (or in its last stretch), I'm looking forward to seeing cool applications. I'm going a bit off track here but relevant to all LD(P) implementations: Technically LD servers are not major hurdles at this point... neither is it the case to get LDN/ActivityPub, WebAnnotation, Memento, and a number of other specs implemented. That's relatively trivial to: 1) having a stable *global* authentication mechanism 2) properly implemented across the board 3) with decent browser support and UX across devices is a major challenge - a bottle-neck that's preventing far more advanced stuff happening than what we're used to seeing. We haven't yet nailed that down. Without WebID-* (TLS, OIDC, FIDO?..), users are fundamentally locked into platforms/web spaces which do their own thing. And, we can't expect client applications and servers to continue to implement 50 different ways to sign-in; some working on certain Webspaces but not the others.. Getting non-technologists to create and manage their own profiles somewhere they sufficiently trust and control is next up on the list. At the moment, it is quite involving - to put it mildly - even for those of us that have been working in this vicinity - decentralisation .. federation - for at least a decade. We're relatively okay with authorization - WAC/ACL seems to be sufficient to handle/speak our *initial* needs. We'll find out more once we have an ecosystem of applications doing RW operations across different Webspaces, *and* users being able to switch between applications seamlessly. That's one of the fundamental changes the LD/Web community needs. So, while it is great to see that there are some LD(P) server implementations out there, they are relatively disjoint nodes in the network, especially concerning the applications that are operating over them. The data may be "interlinked" across the space, but the applications are fixed/immobile. And, that effects the shape of data, as well as its potential reuse. The good news... there a number of initiatives (from standards to activism) and developments happening, or have sufficiently matured, and people are enthusiastic about all this. Slowly but surely. -Sarven http://csarven.ca/#i
Received on Monday, 7 May 2018 15:32:59 UTC