Fwd: Re: Review of Linked Data Notifications

Forwarding for the archives.

-------- Forwarded Message --------
Subject: Re: Review of Linked Data Notifications
Date: Tue, 18 Oct 2016 21:42:06 +0900
From: Sarven Capadisli <info@csarven.ca>
To: Fabien Gandon <fabien.gandon@inria.fr>, amy@rhiaro.co.uk

On 2016-10-18 20:35, Fabien Gandon wrote:
> Hi,

Hi, thank you so much for the feedback!

> I just went through the specs and I found it interesting indeed.
> Below are some (naïve) reactions as I read.

This is all very useful and valid, thanks.

> Best regards,
>  a
> “The notification is data intended for the attention of the receiver, for example: a personal message from a friend; a pingback link; a comment on a blog post; an invitation to collaborate; a calendar reminder.”
> The pingback example immediately brings the question of the relation to the WebMention mechanism:
> https://www.w3.org/TR/webmention/

The Social Web Protocols (SWP) document:

http://w3.org/TR/social-web-protocols

also published by the Social Web WG explains the relations between 
various specs including LDN and Webmention. We refer to SWP in our spec.

> “Sends the notification to the Inbox URL by making a POST request, containing the body in JSON-LD or in another serialization acceptable by the server.”
> I guess I know the answer but any good reason for not mentioning conneg here and putting JSON-LD first?

I guess we didn't want to overload the summary with the details. There 
is the Accept-Post possibility which is discussed later on.

> “1.4 Relation to Linked Data Platform”
> May be indicate upfront here the kind of container you use.

Thanks! Updated Editor's Draft.

> “Following discovery, senders MUST deliver notifications through a POST request to the Inbox URL”.
> It could be read as “it is compulsory to send notifications” ; I would say “Following discovery, senders who want to send notifications MUST deliver them through a POST request to the Inbox URL”.

Sounds better, thanks. Updated.

> “3.3.4 Preventing Abuse”
> Is it creating a new risk of DoS and overflow attacks?

No new risk that we are aware of. It is no different than POSTing 
anywhere. Though it is reasonable to question whether a notification can 
throw a constraint validator into a loop.

> I read rapidly but I didn’t see a delete mechanism mentioned (appart from the fact it is in LDP of course) don’t you want consumers (or other authorized agents) to delete things? How do we prevent the Inbox from getting too big?

We thought about it (there is an issue on updating on github actually 
https://github.com/w3c/ldn/issues/14 ), however, we decided that it is 
orthogonal to this spec. We didn't want to respecify the whole of LDP ;)

Inbox "management" felt out of scope. Related to handling a lot of 
notifications is paging, and we do mention that.

-Sarven
http://csarven.ca/#i

Received on Tuesday, 18 October 2016 14:06:17 UTC