I find if you take it as the rule and not the exception it's much easier to
plan. Seems that way lately with Twitter. :-)
FWIW, I know you hate hearing this, but Facebook's API pushes changes into a
beta staging environment every Tuesday, notifies developers of the changes
as they update it, and t
Ain't it a heap of fun to spend one's long weekend answering support
requests from agitated users, due to something you haven't done? LOL
You're not alone in that boat.
Dewald
On Sep 5, 3:10 pm, PJB wrote:
> Why on EARTH must you guys consistently break things before every
> major holiday week
Actually, before yesterday, Twitter's friends_ids/followers_ids calls
were pretty close to being accurate. Now, with this new fix, they are
totally off from what is displayed on Twitter.com. So, whatever "fix"
was made yesterday, has broken the accuracy of the count (where
accuracy == matching
It seems as though you fixed one thing, and broke something else in
the process.
A large number of our users are complaining that the f/f counts (that
we compute from friends_ids() for example) are woefully off (in the
order of several thousand) from the counts displayed on Twitter.com.
Before y
John, thanks for spending time on this. Any chance we can get a lift on the
follow limits for a temporary time so I can catch up a few users that were
affected by this? Or, if you want to do it on a per-user basis I can send
you the names of the users.
Jesse
On Sat, Sep 5, 2009 at 7:55 AM, John
All the time. I have given up on expecting consistency between summary
numbers and the actual detail data.
Twitter has to do so much data replication for performance reasons, I
think it's extremely tough to keep everything in sync.
It's like the discrepancies in search results and site rankings
Hello
I am seeing right now (14:45 UTC) discrepancy between the list of my
followers as reported by my home page at Twitter.com (which is
correct) and the list reported by the API - which seems to be several
days old (it includes several followers that I blocked recently).
Is anyone else seeing
Awesome. Thanks John, and also to everyone who made this happen.
Dewald
On Sep 5, 10:55 am, John Kalucki wrote:
> Thanks to the efforts of many working late into a Friday night, we
> deployed a fix for this issue starting at about 11pm PDT. The fix was
> verified as working in production at abo
Thanks to the efforts of many working late into a Friday night, we
deployed a fix for this issue starting at about 11pm PDT. The fix was
verified as working in production at about 12:05am PDT.
-John Kalucki
http://twitter.com/jkalucki
Services, Twitter Inc.
On Sep 5, 5:36 am, Dewald Pretorius w
Using Twitter4J, and what I've been seeing is last night,
getFollowersIDs(), including suspended. This AM, the filtering of
suspended accounts seems to be working again.
On the other hand, getFollowersStatuses() doesn't seem to be
filtering out the suspended accounts.
pg
On Sep 5, 7:36 am, Dew
John,
Just so we know, is this 5,000 thing going to be fixed over the
weekend, or will we have to wait until Tuesday?
Dewald
On Sep 5, 12:35 am, John Kalucki wrote:
> We're aware of the problem with the following API not returning more
> than 5,000 followers. Apparently this call has recently
Can Twitter remove the following per hour limit for a little bit after they
fix this (at least for whitelisted IPs and/or OAuth)? This has caused us,
and I'm sure many other apps to pre-emptively unfollow people that they were
not supposed to. This is a BIG problem!
I completely agree with Dewald'
Several months ago there were multiple subtle problems with
consistency in following lists. We've since rolled to a new system
that have numerous constraints in place to prevent duplicates and
provide much stronger consistency guarantees.
There is, however, a known issue around pagination. You ma
There have been times when a friend or follower id has been reported
multiple times. Did you check to see if any of the friend or follower ids
that you get back were duplicates?
- h
On Fri, Sep 4, 2009 at 20:17, Dewald Pretorius wrote:
>
> For @dewaldp I get, via the API social graph methods w
We're aware of the problem with the following API not returning more
than 5,000 followers. Apparently this call has recently been
unreliable was often timing out and returning 503s. A change to fix
the 503s limited the results to 5000 followers prematurely. We're
going to get this back to the >500
For @dewaldp I get, via the API social graph methods with paging:
Friends: 764 , Followers: 3,977.
While on my web profile the numbers are: Friends: 747 , Followers:
3,911.
On Sep 4, 11:56 pm, Dewald Pretorius wrote:
> Not only do the social graph calls now suddenly, without any prior
> warning
16 matches
Mail list logo