> I have noticed that the API sometimes returns user ID’s that are out of
> sync with username. I think one case is where a Alice retweets Bob’s tweet,
> and then Bob changes his name to Charlie. When I try to reply to it, it
> doesn’t show up as “in reply to” to original tweet because the reply
> contains �...@bob” instead of �...@charlie”. It would actually get confusing
> because a new user could sign up as Bob and kind of “take over” Charlie’s
> old @mentions that contain �...@bob”. Is this change an attempt to address
> that, by fixing the screen_name->userid mapping at the time a tweet is
> created?
i haven't thought about that.  perhaps we could make that so.  i need to
investigate that.

>  When we post tweets that include @mentions, can we include our own
> entities/user_mentions in our request body, so that Twitter can notify us if
> one of the mentioned screen names has a different userid than what we were
> expecting and/or one of the mentioned screen names is not a valid screen
> name anymore? That would be extremely helpful in dealing with this edge
> case—even if it was subject to some race conditions over a narrow period of
> time.
> again - great idea.  can't guarantee anything, but let me think about it.
 probably not at the beginning, but this is a good idea.

>  entities/user_mentions/screen_name and entities/user_mentions/text are
> redundant. I would rather just pick the text out of the tweet using original
> tweet text indexed by the indices property, to save bandwidth.
in theory i like this.  i'll probably send a follow up e-mail so that i can
solicit other feedback.

Raffi Krikorian
Twitter Platform Team

Reply via email to