- From: Nicolas Chauvat <nicolas.chauvat@logilab.fr>
- Date: Tue, 31 Oct 2023 13:47:39 +0100
- To: Melvin Carvalho <melvincarvalho@gmail.com>
- Cc: David Mason <vid_w3c@zooid.org>, public-solid <public-solid@w3.org>
Hi List, Le Tue, Oct 31, 2023 at 01:37:47AM +0100, Melvin Carvalho a écrit : > Just covering this one point for now. So a big design decision is files vs > database. NSS is files, for example. Say I have an application that records contacts as first name, family name and postal address and another application that records contacts as first name, family name and phone numbers. With documents / LDP, my understanding is that each app will write the first name and family name in two different documents / triple-containers... and there we have two data silos within a single pod. Unless... both apps agree on a common data model and share the same document/triple-container, but then let's talk about that third app that deals with contacts and has its reasons not to use the same data model in whole... and we are back to two data silos within a single pod, one used by two apps and the other by the third app. With a graph database, I would expect to have a single graph where to store everything and that the two data models are just views/parts of the whole graph (possibly with on-the-fly model mapping or inference to share data between similar models). Of course, for it to work in the real world where not everything is public, I need a mechanism to authorize the access to parts of that graph. Using the above arguments, I fail to see why one would want to walk the files path instead of the database one. Could someone explain to me why the files option seems reasonnable and for which use cases ? -- Nicolas Chauvat logilab.fr - services en informatique scientifique et gestion de connaissances
Received on Tuesday, 31 October 2023 12:47:47 UTC