These are chat archives for csarven/ldn

28th
Aug 2016
Sarven Capadisli
@csarven
Aug 28 2016 00:46
We have an issue and a branch on extensibility. We can revive that if need to on handling different types of inboxes. Right now the minimum is just ldp:inbox. Handling different types of containers is a can of worms
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:15
yes
im currently using solid : inbox
which has the advantage of yielding a 200
Sarven Capadisli
@csarven
Aug 28 2016 01:18
Right. That's great. So, we are working on getting ldp:inbox accepted in LDP ns
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:18
great
Sarven Capadisli
@csarven
Aug 28 2016 01:18
And the LDP ns has decent adoption.
What are some of the Solid implementations?
And how many using the solid-terms ns?
How many LDP implementations do you think will add the solid-terms ns?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:20
how many of those 18 actually work
Sarven Capadisli
@csarven
Aug 28 2016 01:20
How many Solid implementations? Can we start there?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:21
no idea they are probably mostly in that list
Sarven Capadisli
@csarven
Aug 28 2016 01:21
Apache Marmotta works
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:21
actually i dont think any solid implementation works
Sarven Capadisli
@csarven
Aug 28 2016 01:21
Virtually there is 2. And only one of those: node-solid-server is the focus right now. Patches made to Gold. AFAIK.
Ok, so why would any existing LDP implementation would want to go out of their way to introduce a new ns?
Just for a single property?
I think the LDP ns is a good way forward.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:22
node solid server doesnt work with websockets from what i can tell
which breaks the whole 'notifications' loop
same with gold
rww play i think may be inactive
Sarven Capadisli
@csarven
Aug 28 2016 01:23
No it doesn't!
You ahve it wrong. Sorry.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:23
which?
Sarven Capadisli
@csarven
Aug 28 2016 01:23
Again, we are not asking receivers to do anything new here.
It is the senders/consumers
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:24
i really dont know
Sarven Capadisli
@csarven
Aug 28 2016 01:24
What do you not know?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:25
What are some of the Solid implementations?
And how many using the solid-terms ns?
How many LDP implementations do you think will add the solid-terms ns?
Sarven Capadisli
@csarven
Aug 28 2016 01:25
Do you not believe me about what's in or intended (in case we got the wording wrong or confusing) to be in LDN?
Ah ok
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:25
the answer to that
i mainly know about my own apps
i cant really speak for others
maybe solid inbox uses solid : inbox
Sarven Capadisli
@csarven
Aug 28 2016 01:27
LDN asking senders/consumers for one single header is a very minor requirement. Solid is asking all sorts of stuff in the HTTP headers. Please don't overlook that. Solid doing that doesn't make it any more "Linked Data" / FYN friendly than LDN. Same goes for LDP. If Solid/LDP are considered to be "Linked Data"-friendly, so is LDN. 100%.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:27
what link headers does ldp use?
but is anyone using BOTH link headers, and document links?
Sarven Capadisli
@csarven
Aug 28 2016 01:28
rel Container / Resource ..?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:28
ah
5.2.1.4 LDP servers exposing LDPCs must advertise their LDP support by exposing a HTTP Link header with a target URI matching the type of container (see below) the server supports, and a link relation type of type (that is, rel="type") in all responses to requests made to the LDPC's HTTP Request-URI.
That's pretty clear to me. That's a MUST
So, if the debate is that putting stuff in the header like that is not very FYN or Linked Data friendly, well, that's a completely different debate to have and a lot of stuff goes in there.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:30
But is anyone doing BOTH?
Sarven Capadisli
@csarven
Aug 28 2016 01:31
This is about the receiver making that announcement.
What LDN is saying is in fact far simpler. We don't require a receiver to do the link header
LDP requires receiver/sender/consumer understand Link header.
LDN requires sender/consumer understand Link header.
What do you think is a simpler design overall?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:34
LDP
I think there was some debate in the WG
long discussions on where to put links, in the headers, or in the body
Sarven Capadisli
@csarven
Aug 28 2016 01:35
You were debating about what's FYN!!!
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:35
sorry im really tired
can barely type
Sarven Capadisli
@csarven
Aug 28 2016 01:35
:)
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:35
ill go to bed
Sarven Capadisli
@csarven
Aug 28 2016 01:35
Just take it easy. Will visit some other time.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:35
i thought you were asking me some questions
Sarven Capadisli
@csarven
Aug 28 2016 01:35
Everything is going to be okay :)
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:35
and i was trying to be helpful
Sarven Capadisli
@csarven
Aug 28 2016 01:36
I appreciate your feedback.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:36
yw
Sarven Capadisli
@csarven
Aug 28 2016 01:36
I agree with your points fundamentally, but I don't feel comfortable about 1) them being fair in comparison to related work out there 2) being exclusive
So, if Solid/LDP were completely different beasts as far as their requirements from senders/consumers, we'll have something to talk about regarding FYN/LinkedData-friendly or not. However, the arguments made against LDN equally apply to them and other implementations in fact.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:49
looking at related work is useful
comparison is very helpful
the question that comes to my mind is whether inbox is protocol layer or data layer
Sarven Capadisli
@csarven
Aug 28 2016 01:55
Is rel=meta protocol layer?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:56
idk
i think it needs to be specced out a bit more
i know webmention declared itself as a protocol to avoid having to use json
ie to get round the charter
Sarven Capadisli
@csarven
Aug 28 2016 01:58
headers for an image reveal rel=meta in Link. Is the pointer of that best fit for the protocol layer or the data layer?
Solid allows that for all resources and generally considered to be useful for non-RDF resources.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:59
i definitely have need for inbox at the data layer
that's certain
Sarven Capadisli
@csarven
Aug 28 2016 01:59
We agree. We all do see that need.
Melvin Carvalho
@melvincarvalho
Aug 28 2016 01:59
i am unsure i have need for .meta at the data layer
if it's in the data layer i dont think it should be in the protocol layer, and vice versa
Sarven Capadisli
@csarven
Aug 28 2016 02:00
Whether you need that or not is orthogonal to whether it is done "correctly". Assume for a moment that you do need that, does that imply that the approach is FYN/LinkedData-friendly?
Melvin Carvalho
@melvincarvalho
Aug 28 2016 02:00
sorry too tired to parse that, gnight
Sarven Capadisli
@csarven
Aug 28 2016 02:01
I ack your argument. Fair enough to look at it that way too. And we should keep that in mind. However, there are multiple variables here and we need to consider all of them as a whole and arrive at something that maximises the final result
Night