On 2 January 2013 17:26, Mikael Nordfeldth
<address@hidden> wrote:
On ons 2 jan 2013 16:14:55, Melvin Carvalho <
address@hidden> wrote:
> -1
>
> Activiity streams is a neat idea, and a few years ago when the protocol
> started it was promising in that it was an early embracer of the JSON
> paradigm. JSON, imho, is an excellent choice for developers in that you
> can easily serialize it into most languages.
Given that the very promising networks surrounding XMPP would interact splendidly with properly namespaced XML schemas (i.e. activitystreams as used in BuddyCloud etc.), it's much better for interoperability than any JSON namespaceless serialization.
OK, I would suggest you're in a minority there.
> Today there are several first class
> JSON serializations, better engineered, with better interop and with
> better adoption.
I'd like to cite the "running code" part of http://en.wikipedia.org/wiki/Rough_consensus and quickly end this abstract discussion on would-be utopian serialization standards.
It's all just ones and zeros. If you're proposing something truly awesome, like ternary computing protocols I might be persuaded into discussing a possible migration. I doubt anyone else would, though.
Not sure I understand the point you are making. I agree that digital code is binary. But not all binary formats are designed to interoperate. Activity streams does NOT have a good track record of interoperating with something other than itself. It is the 'one serialization to rule them all' argument. Linked data DOES have a good track record of interop. It's also doubling every 9 months, whereas activity streams projects seem to be dying. Case in point
status.net is shutting down to new users.
> There is also the issue of the proprietary nature of activity streams. I
> would be much happier if they were to put it under creative commons. It
> is the only spec that I know where it has been delayed in order to get
> approval form lawyers.
APIs and similar specs are AFAIK not subject to copyrights or patents, following the result of lawsuits like Oracle vs. Google on the Dalvik engine.
I would like to understand this more. Do you have a pointer.
Do you think you could persuade activity streams to come under a non proprietary license?
> I dont want to be too unkind to AS as my name is on the spec :) But I
> would consider it one of the weaker serializations that should be
> supported as legacy, if there is a very compelling reason to do so.
Are there any inherent restrictions that will make it [practically] impossible to use activitystreams in future federations with other software? Will our social networking change so radically that we no longer "share" or "comment" stuff that have one or several objects possibly containing other objects?
Sure
If so, how far into the future is that? And would it be easier to migrate from GNU Social's current homebrew structure - or something properly specified like ActivityStreams?