Re: Getting the group back on track

On 10/16/2015 01:41 PM, Melvin Carvalho wrote:
> On 16 October 2015 at 13:24, elf Pavlik <perpetual-tripper@wwelves.org>
> wrote:
> 
>> On 10/16/2015 12:59 PM, Melvin Carvalho wrote:
>>> On 15 October 2015 at 18:58, elf Pavlik <perpetual-tripper@wwelves.org>
>>> wrote:
>>>
>>>> On 10/14/2015 09:13 PM, Jason Robinson wrote:
>>>>> Hey,
>>>>>
>>>>> A big +1 the the email of Christopher.
>>>>>
>>>>>    > Right now, off top of my head implementers would be:
>>>>>    >
>>>>>    > 1) IBM Connections
>>>>>    > 2) Pump.io
>>>>>    > 3) MediaGoblin
>>>>>    > 4) Objective8 (Thoughtworks)
>>>>>    >
>>>>>    > Anyone else?
>>>>>
>>>>> I'm pretty sure if a clear and well thought JSON based language comes
>>>>> out, diaspora* is interested. However, AS2 is only the language part,
>>>>> personally waiting to see how the protocol parts start to take shape.
>>>>> Hoping to have more time to contribute to those stages which imho are
>>>>> more important for our use case at least. TBH, the way that objects and
>>>>> actions are presented for transfer is only a minor part in the whole
>> big
>>>>> engine of two servers exchanging messages. For diaspora*, and the
>>>>> Friendica + Hubzilla that are connected with it using the same
>> protocol,
>>>>> the server to server is key.
>>>>>
>>>>> Personally I hope the language semantics part could be quickly locked
>>>>> down and the work moved on to figure out the protocol stuff. There is
>>>>> not going to be a "works for all final" version delivered by any group
>>>>> ever - everything is always iterated on, and it is better to deliver
>>>>> something concise and small first, instead of trying to tackle
>>>> everything.
>>>>>
>>>>> Just a few comments regarding how diaspora* federates. Basically we
>> have;
>>>>>
>>>>> * webfinger and .well-known/meta-info for discovery
>>>>> * XML language for actions and content
>>>>> * Salmon Magic Envelope for signing the XML content
>>>>>
>>>>> The project is in the works of pushing out the federation code to a
>>>>> separate repository, which means it would be easier to start using
>>>>> another protocol in some future. The key things that we need however
>> are
>>>>> pretty much the three items above;
>>>>>
>>>>> * discovering
>>>>> * describing content
>>>>> * authoring
>>>>>
>>>>> Especially the last one is something that I'd be interested in hearing
>>>>> some thoughts about, what kind of idea has this group got on how to
>> sign
>>>>> AS2 JSON content payloads? Outside diaspora*, I've got some personal
>>>>> plans on creating a Python library to abstract several protocols, as an
>>>>> experiment if nothing else. Currently it supports diaspora* for some
>>>>> limited stuff, receiving and sending posts, and I would like to add
>> some
>>>>> AS2 based routes there too. Content signing is *the* most important
>>>>> thing to get right.
>>>>
>>>> We had months ago bit intense conversation around JSON-LD Signatures
>>>> * http://manu.sporny.org/2013/sm-vs-jose/
>>>> * https://youtu.be/QdUZaYeQblY
>>>> * https://github.com/digitalbazaar/jsonld-signatures
>>>>
>>>> I hope to give them a try in near future. Also combined with content
>>>> addressable versioning of documents...
>>>>
>>>
>>> Thanks for the pointers.  I use this too.  I first became interested in
>>> this back in 2001 when I was the lead developer for implementing digital
>>> signatures at Deutsche Bank.  It emerged that there were problems with
>>> standard document signing and signing of XML which was problematic in the
>>> finance world.  Since that time I've been searching for a solution to
>> this
>>> problem and LD signatures solves pretty much all problems out there.  Do
>>> note that it was designed for high value transactions, so may be overkill
>>> for more casual aspects of the social web.
>>
>> Melvin, as I understand LD Signatures rely on RDF Dataset Normalization
>> * http://json-ld.github.io/normalization/spec/
> 
> 
> yes, json-ld signatures (jsigs) now has the latest 2015 graph normalization
> and signature algorithm implemented. Brian Sletten just implemented the
> same in the rdflib Ruby library too. Python and PHP implementations are
> being updated
> 
> note these are updates as of the last week
> 
> 
>>
>>
>> and while work with JSON-LD in very elegant way, they don't stay JSON or
>> XML or RDFa or Turtle specific? One can use them in the same way with
>> any of RDF serializations?
>> * http://www.w3.org/TR/rdf11-new/#section-serializations
>>
>>
>>
> exactly, it can be used with any data, and any serialization ... most
> recently I am finding turtle more attractive as it offers advantages, but
> the same system works with XML or JSON LD too
> 
Thank you!

I believe this addresses Harry's concern about *yet another JSON signing
spec*. Once again, while LD Signatures work in elegant way with JSON-LD
and even already provide .js library to use them. People who prefer to
work with Turle, RDFa etc. can still use the same technology. I
understand that XML, Turtle and HTML based serializations stay *out of
scope* for this group, but people don't work on LD Signatures within
charter of this WG.

Received on Friday, 16 October 2015 12:06:05 UTC