Now I'm testing a new feature: additional User attributes shown in a "Users of message" view: numbers of posts, friends, followers and favorites. And it appears that different GNU Social instances have very different information about the same User as if each of them counts only what it knows about. E.g. about this my account andstatus@loadaverage.org : Loadaverage thinks that it has 241 followers while Quitter.se - 67 only. Very interesting. On this feature see https://github.com/andstatus/andstatus/issues/320 @mmn @vinzv
@andstatus Yep, !qvitter solves this by requesting the data directly from the origin server of the account (when you hover the mouse over a profile a request is sent to the remote server).
@andstatus It's only the Twitter API that doesn't have that. And global _identification_ wouldn't help getting a full picture of the user profile.
Besides, I don't think it would be safe to expect that a remote site is even willing to provide full info on how many posts someone's made (why compete in spammyness?!) and that information could also be extremely skewed if an instance for example decides to auto-delete old posts etc...
@mmn Global User's identification will help clients to create broader/more complete/objective view on a User and his activity based on information from several GNU Social instances (where that client has accounts...)
@andstatus As I mentioned, there is already a global user identifier. The URI for users will always be the same across profiles. And profile URLs all have metadata for various methods of discovery of this URI (usually it ends up being the /user/$id URL). If that data is not part of the Twitter API then we can arbitrarily add it to ease API usage. But the Twitter API sucks and the ActivityStreams (.atom and .as endpoints) are the ones that give proper, federation-usable data.
@mmn AndStatus already supports ActivityStreams protocol in JSON format. Do you mean that I can switch to that protocol accessing GNU Social and forget about Twitter API problems? If yes, do you have any links with a list of API calls... ? Maybe any reference client implementation?!
@mmn As you didn't answer yet, I will put my question another way: Do you, as a main GNU Social developer, recommend to switch a client application from Twitter API to Activity Streams in JSON format? The format specification got a follow up, which we may become confirming to in order to ease integration with other systems: https://www.w3.org/TR/activitystreams-core/ ?! BTW congratulations for making more commits to GNU Social codebase than Evan did. PS: I suspect that no GNU Social client application uses Activity Streams JSON format to connect to GNU Social. Am I right?
@mmn I understand your no reply as a sign that I should not plan to switch AndStatus as a GNU Social client application from Twitter API to Activity Streams in JSON format?!
@andstatus Sorry I didn't answer because I didn't have time and then forgot :) I think it would be better to use the AS json data than the Twitter json output! It should have the same information and more of it too!
So far I've relied on that data is properly converted from the internal data format (which I make sure works in the AS in Atom output) to AS in JSON. So far I haven't broken anything to my knowledge, but since it is less used overall there is a higher likelihood of unnoticed issues.
@mmn Thank you for encouragement. Do you know any implementation of this API that I could lookup in order to move from a zero point?! I need at least two pieces: authentication (client registration...) and making a request something like "statusnet/config"