- From: Harry Halpin <hhalpin@w3.org>
- Date: Fri, 23 Oct 2015 16:51:55 -0400
- To: Aaron Parecki <aaron@parecki.com>, Amy G <amy@rhiaro.co.uk>
- CC: Social Web Working Group <public-socialweb@w3.org>
- Message-ID: <562A9DEB.50100@w3.org>
On 10/23/2015 01:06 PM, Aaron Parecki wrote: > It doesn't cause any problems *parsing*, but it means you can't use > object notation in most languages. Apologies, I meant "parsing into object notation." I'm not quite sure why "@" was chosen but it does have this problem with almost all JSON-languages I know, and so using @context, @language, @type, etc. all requires very special handling, which may limit uptake. cheers, harry > > $foo->@context is not valid syntax in PHP or nearly any language I can > think of. It limits the use to array/hash notation, like > $foo['@context'] in PHP. > > ---- > Aaron Parecki > aaronparecki.com <http://aaronparecki.com> > @aaronpk <http://twitter.com/aaronpk> > > > On Fri, Oct 23, 2015 at 10:03 AM, Amy G <amy@rhiaro.co.uk > <mailto:amy@rhiaro.co.uk>> wrote: > > The '@' symbol seems not to cause any problems with default json > parsing in python and php, what were you using? > > Amy > > On Oct 23, 2015 9:14 AM, "Harry Halpin" <hhalpin@w3.org > <mailto:hhalpin@w3.org>> wrote: > > Elf, > > On 10/23/2015 04:49 AM, elf Pavlik wrote: > > On 10/22/2015 06:03 PM, James M Snell wrote: > >> On Thu, Oct 22, 2015 at 8:51 AM, Harry Halpin > <hhalpin@w3.org <mailto:hhalpin@w3.org>> wrote: > >> [snip] > >>> I'll try to get to this next week, but my high-level > feedback is likely > >>> for AS2.0 to be successful everything outside the basic > actor-verb model > >>> and the kinds of metadata in Winer's RSS specs/Atom should > be removed > >>> and put back in Activity Vocabulary. > > Harry, can you please reply with a short snippet of AS2.0 > JSON data > > which uses at least one term defined in microformats.org > <http://microformats.org> vocabulary? > > > > I believe that you feel comfortable with backing you > proposal with > > simple 5 lines of plain JSON which I see you keep promising > to people! > > > People who send email use natural language :) > > > > > > > >> This makes no sense given that all of the properties are > ALREADY > >> defined in the Activity Vocabulary. The Core spec deals > only with the > >> serialization and relies on the Vocabulary document to > define the > >> actual terms. > >> > >>> I also am still strongly against the Activity Vocabulary > being a > >>> normative Recommendation, as it will lead to endless > bikeshedding and > >>> its a Sisyphean task to describe all social interactions > using a single > >>> vocabulary, and the vocabulary should align where possible > with > >>> IETF/microformats specs down to the 'string' level. > >>> [snip] > >> The Vocabulary does not attempt to define all social > interactions, > >> just a handful of those that we know are already relevant > to a good > >> number of existing social systems. If there are suggestions for > >> removing specific terms, then I'm all for looking at those. > >> > >> As for bike shedding, if the minimal set of terms defined > in the > >> vocabulary document are not to any specific implementers > liking, there > >> is a well defined extensibility mechanism that allows > developers to > >> use terms from other vocabularies quite easily. > Implementing support > >> for such extensions is fairly trivial (e.g. > >> https://github.com/jasnell/as2-schema) > > Does this well defined mechanism still work if > implementation chooses to > > ignore JSON-LD context? I keep hearing from Harry about > intentions for > > such practice becoming common and I would like to verify > that we don't > > contradict ourselves here! > > Again, due to a relatively simple spec error on the part of > the JSON-LD > editors/Working Group, @context and any other attribute > defined with a > '@' symbol are not processed out of the box as objects by most > modern > programming languages. Thus, you have to give any JSON-LD > defined '@' > symbol special processing. While there it is possible everyone > will > start using JSON-LD libraries, I expect many if not most > developers > will not use JSON-LD libraries but will want to consume AS2.0 > as JSON. > It's possible I'm wrong, but that's the feedback I've gotten from > Thoughtworks (whose IE application is still waiting) and others. > > In other words, we need to keep JSON-LD to keep RDF interop, > but realize > most people are not using RDF-based programming stacks. If > AS2.0 is to > be a genuine interop layer, design needs to take that into > account and > if JSON LD conventions are broken, c'est la vie. > > cheers, > harry > > > > > >> At this point in the process, it would be far more > productive to focus > >> on implementation and fixing the specific parts of the spec > that make > >> implementation difficult, etc. > >> > >> - James > >> > > > >
Received on Friday, 23 October 2015 20:52:00 UTC