Re: Apologies (former: ActivityStreams Schema: Hierarchy of Types)

hello elf.

On 2014-11-13, 12:38, ☮ elf Pavlik ☮ wrote:
> On 11/13/2014 12:51 AM, Erik Wilde wrote:
> please accept my apologies for possibly turning our conversation
> yesterday into an argument :(

there's absolutely no need to apologize. you raised a valid question and 
i was just thinking about a good way to respond. i should do that with 
actual examples, but before i do that, here are the some things we 
should care and think about:

- cases in which a client with JSON goggles sees something, and clients 
with JSON-LD goggles see something else. in the end, AS should be clear 
about what the relevant content of AS is, and how clients are expected 
to extract that from AS. if we have cases where that's not true, we have 
a problem, because different implementations then see different things, 
and that's an interoperability problem.

- cases in which a client with JSON-LD goggles sees one thing, and this 
can be serialized in ways that lead to different JSON "views" of that 
thing. (a classical example for this kind of problem are XML namespace 
prefixes, which are supposed to be irrelevant, and to complicate things 
further, not all implementations allow full control over serialization.) 
this would mean that JSON-LD-goggled clients would never realize nor 
might they even have control over the fact that some JSON-LD 
peculiarities can result in relevant differences for JSON-goggled clients.

- another thing i was wondering about is context drift. if the JSON 
counts, then the JSON-LD interpretation may change if the context 
definition changes, and that's neither under control nor even visible to 
AS producers and consumers. that's for example one reason why in many 
communities, any external information that materially changes a 
document's interpretation (classical examples are schema languages with 
default values) is severely frowned upon. is that a scenario we want to 
consider? should we at least mention it and say that JSON and JSON-LD 
interpretation can drift apart in the light of context changes?

but you were absolutely right to ask for examples, and that's on my to 
do list. i hope my (still generic) explanations are a little bit clearer 
with regard to the scenarios i am worried about.

thanks and cheers,


erik wilde |  -  tel:+1-510-2061079 |
            | UC Berkeley  -  School of Information (ISchool) |
            | |

Received on Thursday, 13 November 2014 16:10:14 UTC