Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Aug 11 20:52
    @RubenVerborgh banned @mikeadams1
  • Jan 04 20:23
    @RubenVerborgh banned @WebCivics_twitter
  • Jan 04 20:18
    @RubenVerborgh banned @SailingDigital_twitter
  • May 27 2019 06:08
    User @Mitzi-Laszlo unbanned @in1t3r
  • May 23 2019 06:49
    @Mitzi-Laszlo banned @in1t3r
  • May 16 2019 09:49
    @Mitzi-Laszlo banned @mediaprophet
  • Feb 01 2019 22:04
    User @melvincarvalho unbanned @namedgraph_twitter
  • Feb 01 2019 21:49
    @melvincarvalho banned @namedgraph_twitter
Vincent
@Vinnl_gitlab

@csarven Some use cases are a bit hard to describe in that format. For example, I have the use case "I want to be able to allow one person to see but not modify a single document", which is technically possible with the existing WAC system, but suddenly becomes really hard (and thus comes with risks like privilege escalation) if an access rule already exists mentioning that person and that document in addition to other persons and documents (see solid/authorization-panel#75).

How can we indicate that it would be nice if we could do the above by removing/writing only the relevant Triples, without that affecting other access. Is that even something you're looking for with this doc?

Benoît Alessandroni
@balessan
@michielbdejong : do you have some implementation examples somewhere of that indexing ? And some references to the private type index
We are not using that no :-)
Fred Gibson
@gibsonf1
@michielbdejong We're very interested in the intent behind the public and private type indexes too - is there any documentation on that or info to share?
Vincent
@Vinnl_gitlab
Fred Gibson
@gibsonf1
Thanks @Vinnl_gitlab - very helpful! For the naming of things: /settings/publicTypeIndex.ttl does the .ttl extension have to be there? (In our case, it's just another node in the graph like any other, not a .ttl file)
Tim Berners-Lee
@timbl
@balessan Querying across pods and across a server is on the Solid roadmap. https://solidos.solid.community/public/Roadmap/Tasks/state.ttl#Iss1595438207094
Fred Gibson
@gibsonf1
@timbl is there any plan for finding webids of people regardless of which server they may have their pod on - some kind of directory that all servers could access and update?
Tim Berners-Lee
@timbl
It’s in a state that we have committed to do it at some point but are not working on it at the moment. There has been talk of SPARQL and Fragments for instance. Are you interested in helping define it?
Fred Gibson
@gibsonf1
We will need it soon, so I would be happy to help
In our case, we have pods for people, companies, facilities, software systems, etc - so the indexing could also include type of entity the pod represents
Tim Berners-Lee
@timbl
There was one, called IIRC webizen.org whoever started it i think stopped maintaining it. Anyone remember who that was?
Fred Gibson
@gibsonf1
Maybe there could be a websocket server to server communication about adding/changing pods, and each server could maintain the overall index locally as it changes
and included there would be a websocket command, for a new server starting up, that would download all the pod index information from all pod server to get in sync
for each pod, you would only need name, type, webid and maybe location information if public
Fred Gibson
@gibsonf1
We currently have a pod index pod on each pod server that indexes all the pods on that server
Fred Gibson
@gibsonf1
@Vinnl_gitlab For the schema of the public/private type index, would it be something like a container for each type of item indexed such that members of that type in the pod are automatically added to the container?
Vincent
@Vinnl_gitlab
@gibsonf1 I don't have much more information than is on the page I linked, but as I understand it, data of a given type can either be collected in a Container (and referred to from a type index using solid:instanceContainer), or in a specific document (using solid:instance). But nothing would happen automatically; it's up to client applications respect listings in the type index, and to add their own if no listing for their data exists yet.
Benoît Alessandroni
@balessan
Thks @timbl good to know those questions are on the roadmap :-)
thanks @Vinnl_gitlab I will check the resource.
Mark Hughes (happybeing/theWebalyst)
@happybeing

Due to a disagreement with Twitter on whether they should have my phone number on top of my real name, I'm revisiting mastodon, so if any of you folks are also using mastodon and wish to connect, you'll find my tech account here: https://mastodon.cloud/@SAFEpress

Mastodon has improved a lot (mainly in responsiveness and quality of the community) in three years, and there's a very nice alternative UI called Pinafore, and an associated side-bar add-on for Firefox.

Hopefully Solid can move into this space too before long. I think comparing any Solid offerings with the performance of mastodon will be a good benchmark - it is seriously responsive. I have two accounts on different instances (servers) and if I have both open and toot or delete toot on one, the feed on the other updates instantly. Same with follows/unfollows, it is impressive and a far nicer UX than twitter. I think Solid should aim to match the mastodon UX in terms of usability and responsiveness, and if it does this could help adoption as I think twitter is getting less friendly and more invasive (as to be expected).

So join me on mastodon for a play and a chat. Lots of like minded folks on there. https://mastodon.cloud/@SAFEpress

Fred Gibson
@gibsonf1
@Vinnl_gitlab One thing we are doing with TrinPod is that we have what we call a OneWorld model for each pod in which all information is modeled as entities and events that are conceptually linked and change through time, and containers are simple indexes of that model. So the source of truth for the entity/event is not in the container. So in that context, automated containers make sense, but if source of truth are files in containers, it's not really possible to do it.
matrixbot
@matrixbot
TomasEkeli yes, mastodon has come far - also other open platforms like e.g. peertube and element (née matrix) - what's the relationship between solid and these?
huhn511
@huhn511
Hey, I'm searching for helping hands and minds for a rust solid server implementation. Already started with a friend a few weeks ago. If you want to help, please write me :) Thanks :v:
7 replies
tsojcanth
@tsojcanth
I have a silly question, but I was wondering about a couple of things. Does solid write any metadata about who creates or touches a file?
Michiel de Jong
@michielbdejong
@tsojcanth not server-side. but we often do so client-side, so just include dc:author triples etc into your data
tsojcanth
@tsojcanth
@michielbdejong thanks
other newbie question: we have a necessity of implementing blacklists. As far as I'm aware it's not possible to do that through solid directly, nor at the SSL offload level by looking into headers. do you have any other suggestion, beside throttling/blocking IPs?
append-only inbox folders are great but it's easy to DOS pods by filling them with garbage
Michiel de Jong
@michielbdejong
right
you should never trust the data inside an inbox item by itself (unless it's signed with a W3C-VC or similar)
that means you would always look up the URL that's mentioned in the inbox item
and from there you can verify the sender
tsojcanth
@tsojcanth
it's not a matter of trusting it, it's a matter of allowing attackers to write in it
Michiel de Jong
@michielbdejong
if the sender is not in the user's addressbook, nor a friend of a friend, then you could just ignore the inbox item
ah you're worried about file size?
tsojcanth
@tsojcanth
i'm worried about Alice filling Bob's pod with garbage so Bob can't get Charlie's mail
Michiel de Jong
@michielbdejong
i thought you were thinking more at the level of the user having to sift through annoying lists of spam
why would Bob not get Charlie's mail? would it not arrive?
tsojcanth
@tsojcanth
because the pod is full
Michiel de Jong
@michielbdejong
ah
so file size, yes
tsojcanth
@tsojcanth
not only file size, also file numbers
Michiel de Jong
@michielbdejong
right, so an enemy of Bob whose aim is to annoy him
or block his work
tsojcanth
@tsojcanth
(it's easy to filter when listing! but beside running an agent cleaning folders I have found no way to do it automatically)
Michiel de Jong
@michielbdejong
yeah, then you should look at throttling and blacklisting probably. maybe you could use CloudFlare?
i personally think from my experience as an email user, spam is a bigger problem than dos
tsojcanth
@tsojcanth
yeah, throttling and blacklisting IP is not really good enough, but spam is easier to handle as you can blacklist as you process your inbox. there could be an agent running the blacklist for you, but that involves scheduling tasks somewhere
if there were a header with the user identity (in conjuction with acl:AuthenticatedAgent rather than merely public) it would be a solution.
Michiel de Jong
@michielbdejong
@tsojcanth then you could still get millions of inbox message, they would just come from as many randomly generated webid's