[twitter-dev] Site Streams DMs missing?
One of my test users was not getting DMs for a few hours yesterday (others too I learned) via Site Streams. This happened about noon EST yesterday. Here is the exact circumstance: 1. Both user id '123' and user id '456' are under my control. 2. User 123 sends a DM to user 456. 3. The stream gets only one DM. It is the message with for_user 123 and recipient_id 456. This shows the DM sent. 4. The DM showing for_user 456 and recipient_id 456 did not occur. At around 12:30 EST or so I noticed that both messages were in the stream and all was well. I have not retested today yet. But I plan to. Mark -- Have you visited the Developer Discussions feature on https://dev.twitter.com/discussions yet? Twitter developer links: Documentation and resources: https://dev.twitter.com/docs API updates via Twitter: https://twitter.com/twitterapi Unsubscribe or change your group membership settings: http://groups.google.com/group/twitter-development-talk/subscribe
[twitter-dev] Re: Site Streams
Taylor, Our user base for this is very small, only a tiny fraction selected for the test... Mark On Jul 6, 9:47 am, Taylor Singletary wrote: > Orian, > > It's just that one shouldn't go to production without checking in with us > first right now so we can make sure the capacity outlook is good. We're > getting closer to increased capacity on Site Streams, but until then we're > cautious with additional production consumers, especially with larger user > bases. > > @episod <http://twitter.com/intent/user?screen_name=episod> - Taylor > Singletary > > On Tue, Jul 5, 2011 at 10:39 PM, Orian Marx wrote: > > Mark, this is a great post, thanks for taking the time to write it up. > > What I'm curious about is that I believe Site Streams is still in beta > > and not supposed to be used in a production environment according to > > Twitter. Has it been stable for you? > > > On Jun 28, 12:24 pm, Mark Krieger wrote: > > > We've gone live with Site Streams for paying customers of our > > > TweetRoost product. I wrote a blog with technical details which some > > > people might find interesting. See it athttp:// > >www.mediaroost.com/2011/06/tweetroost-goes-live-with-twitter-s... > > > -- it is not a product blog or a pitch :) (and I am a big fan of Site > > > Streams, it was a great project for us to implement) > > > > Mark Krieger > > > President > > > @mediaroost > > > -- > > Twitter developer documentation and resources:https://dev.twitter.com/doc > > API updates via Twitter:https://twitter.com/twitterapi > > Issues/Enhancements Tracker: > >https://code.google.com/p/twitter-api/issues/list > > Change your membership to this group: > >https://groups.google.com/forum/#!forum/twitter-development-talk -- Twitter developer documentation and resources: https://dev.twitter.com/doc API updates via Twitter: https://twitter.com/twitterapi Issues/Enhancements Tracker: https://code.google.com/p/twitter-api/issues/list Change your membership to this group: https://groups.google.com/forum/#!forum/twitter-development-talk
[twitter-dev] Re: Site Streams
We are doing this in a VERY controlled way, where the switch from REST API to Site Streams (and back) is very simple as needed in our application. Also, the users who get Site Streams are VERY few, and are aware of it. So far, so good, we are doing very well with it. It has been up for many days, no interruptions. Happy happy. Mark On Jul 6, 1:39 am, Orian Marx wrote: > Mark, this is a great post, thanks for taking the time to write it up. > What I'm curious about is that I believe Site Streams is still in beta > and not supposed to be used in a production environment according to > Twitter. Has it been stable for you? > > On Jun 28, 12:24 pm, Mark Krieger wrote: > > > We've gone live with Site Streams for paying customers of our > > TweetRoost product. I wrote a blog with technical details which some > > people might find interesting. See it > > athttp://www.mediaroost.com/2011/06/tweetroost-goes-live-with-twitter-s... > > -- it is not a product blog or a pitch :) (and I am a big fan of Site > > Streams, it was a great project for us to implement) > > > Mark Krieger > > President > > @mediaroost -- Twitter developer documentation and resources: https://dev.twitter.com/doc API updates via Twitter: https://twitter.com/twitterapi Issues/Enhancements Tracker: https://code.google.com/p/twitter-api/issues/list Change your membership to this group: https://groups.google.com/forum/#!forum/twitter-development-talk
[twitter-dev] Site Streams
We've gone live with Site Streams for paying customers of our TweetRoost product. I wrote a blog with technical details which some people might find interesting. See it at http://www.mediaroost.com/2011/06/tweetroost-goes-live-with-twitter-site-streams/ -- it is not a product blog or a pitch :) (and I am a big fan of Site Streams, it was a great project for us to implement) Mark Krieger President @mediaroost -- Twitter developer documentation and resources: https://dev.twitter.com/doc API updates via Twitter: https://twitter.com/twitterapi Issues/Enhancements Tracker: https://code.google.com/p/twitter-api/issues/list Change your membership to this group: https://groups.google.com/forum/#!forum/twitter-development-talk
[twitter-dev] Re: A new permission level
I could not agree more! I was about to write the same when I saw this message. I do not mind changing the Read/Write on my apps page. I do not even mind having to change my app to allow 'Re-connect' to re-authenticate. Even though it is a pain (but it is done at least). I DO mind now having to live with many many users not reading our mail about having to re-authenticate and then being angry with my company. They'll even be angry at us and confused that they have to re-auth if they do read the email, since they knew what they were doing when they auth'd in the first place. Twitter: Please grandfather in old users who've accepted that their DMs are being sent and received by our apps. It will save us and our users from a lot of headache. Thanks, Mark Krieger, Mediaroost @mediaroost, home of TweetRoost On May 18, 4:27 pm, Dewald Pretorius wrote: > The more I think about this, the less it makes any sense whatsoever to > force everyone through a re-authentication if DM access is required. > > Here's why: > > 1) For existing user tokens, the users have already granted access > with the knowledge that it is to their DMs as well. In other words, > they have already granted access to their DMs. > > 2) If an app needs access to the users' DMs, it is going to force > thousands of people to waste thousands of hours to re-authorize > something they want the app to do and something they have already > implicitly granted to the app. > > 3) Many users are going to miss the memo, and then be very upset with > the app owner(s) because what had worked before suddenly stopped > working. > > 4) Additional and completely unnecessary workload and costs are going > to be added to the support staff of the app, to help users who do not > understand why they need to re-authorize, or who have missed the memo > in the first place. > > 5) By forcing re-authorization for apps that require DM access and > already have DM access, Twitter gains absolutely nothing. After > forcing thousands of people through a redundant process, we're back at > where we started, namely, the app has access to the user's DMs. It's > not like the user has a choice of not granting a requesting app access > to his DMs, but only to his followers and tweets. If the app request > DM access, the user can either grant it, or deny access completely. > Exactly the same way it works today. > > The only benefit here is for apps who don't need DM access, which will > now be able to request account access without DM access. But, if the > app does not need or use access to DMs, it provides absolutely no > benefit to take existing DM access of already granted user tokens > away. It is not used. > > It makes perfect sense to implement this change from a date going > forward, meaning all user tokens granted after that date will be > either Read, Read & Write, or Read & Write & DM. That provides more > transparency for the user. But to yank away existing access rights and > then force the equivalent of a small nation through a re- > authentication process just to re-establish what had already been > granted and then unilaterally taken away, that makes no sense at all. -- Twitter developer documentation and resources: https://dev.twitter.com/doc API updates via Twitter: https://twitter.com/twitterapi Issues/Enhancements Tracker: https://code.google.com/p/twitter-api/issues/list Change your membership to this group: https://groups.google.com/forum/#!forum/twitter-development-talk
[twitter-dev] Complex bug?
We have Site Streams pretty much working here in Rooster-land, need to do some cron cleanup, but everything is looking very happy, the Site Streams and the Rest API windows have been running side by side for quite a while, all looking very 'the same' yay. Thanks Taylor, this is wonderful, and your help with our one or two problems has been great. And very appreciated. We will keep freemiums likely on Rest API, for a lot of reasons, at least for now, but I did find a nasty bug, not sure if someone reported it before, and it was not hard for us to track down but we were lucky, I wanted to report it: A customer clicked for Mentions, we used Rest API to get them, but he got an error screen, it was strange, the data in the API record was a mess, according to the error, and we wrote the data we got to a database so we could do a postmortem. Both the error messages and the db show fields missing in the API data. (Not zero or null, but MISSING, as in, the array indices for the fields do not exist when parsing the json). Fortunately, the tweet data itself was there, and it had a bit.ly link (good luck) so we could track down the record, and it was deleted -- I looked in Twitter itself, the id is gone --- I am guessing the delete was done about the time the API request was made. So the record is in the process of being deleted, Rest API is called, a partial mal-formed record gets sent to my app, it is unhappy. We could put in protections, but we would need to sure protect a LOT of code from this. Is this a known problem? If not, I could provide a lot of data, and do it offline to not pollute the group if this would be helpful. And thanks again, Taylor for your help! Mark -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] Re: User mute, sounds reasonable?
Our external application does this, it is a breeze, check out my blog at http://bit.ly/kD0Je5 On Apr 2, 12:00 am, sromero wrote: > I would like to what you think about integrating to the web interface > a "mute" functionality. Yes, it would be similar to follow/unfollow at > the end, but I think that having a small list somewhere (probably at > the right of the time line right now) where I could "unmute" users > with a single click without having to remember who I have muted > (therefore having to look for them again, sometimes with the need of > writing down the username first) can be useful. Examples: > > - You may want to mute some journalists commenting a match I don't > care > - You may want to mute certain people/TvChannels/etc during your work > hours > > I say it's similar to follow/unfollow because I think it would add a > third "state" in the relationship between two users, and thus probably > change part of the API: > > - Not Following > - Following > - - Muted (which can only happen if following) > > The functionality can be done using the current API with an external > application I guess (I haven't gone deep into the API), but we obtain > to lists from the user: following and followers, and thus we don't > know which users are actually muted. That is, we must store this > information in an external place and that might not always be possible/ > desired. > > May be when viewing someone's timeline it could be distinguish those > tweets that happened while you had the user muted (so you know what > you don't remember having read that :) > > Any comments? Do you think that this feature deserves a place in the > API? -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] Site Streams strategy
I have Site Streams implemented for a number of twitter accounts on my testbed system. It is working like a charm, and for this system at least, the home timeline, mentions and messages are all super fast (obviously). It saves API calls, it is nice, I am pretty happy. I wonder: What other data coming down the wire are most likely to save API calls -- and are thus worth the effort to capture in the local fast database for async processing? Other than tweets (including mentions) and messages (DMs) thanks, Mark -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] Re: wait time for Site Streams whitelist ?
Taylor, Can you give us some time estimate? I have a development team stuck now, nothing more we can do to test with User Streams. I put in weeks ago too for this whitelisting. Mark On Apr 5, 7:46 am, Nicholas Chase wrote: > I put in my request on 2/21 and got approved on 3/3. I should point > out, though, that I did email (just once!) to follow up. > > Nick > > On 4/5/2011 7:02 AM, David W wrote: > > > I'm still waiting too. > > > Put the request in on 28th Feb. Got a response to sign ToS just under > > 3 weeks later. Responded immediately, been waiting over 2 weeks for a > > response to that. -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] Site Streams and lists
While waiting for Beta Site Streams access, I have been testing with User Streams, and reading docs. I have a lot done, it is a neat way to do things. I've easily implemented mentions, dms, hometimelines, and they're working in my product using User Streams as the testbed. Doing Site Streams will be more difficult, but the basics are there for me. I have a question about lists though: I know I can get list changes (I can see them in the stream), but more important to me in principle is getting the List's timeline, since some people use lists extensively and want to see their list timelines and I do not want to waste REST API calls. But it seems to me that not only do List timelines not come down the wire for streams, but it also looks like I cannot 'cobble them' together; the way I might cobble them together for timeline for my followings. That's because people in the List who I follow are not necessarily people who are in my 'regular' followings. Is this correct? Did I miss something? It is probably not horrible, but I just want to make sure I am understanding this correctly. Mark -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] Streams question
We coded our enterprise application anticipating getting whitelisted, but alas that seems to not be in the cards. So we have started to code against the Streams API, but I have a few questions, I hope this is the right place: a. I applied for Site Streams Beta Whitelisting more than a week ago, the application said it is looked at more or less once a week. Is this accurate? When might I hear? (Company is MediaRoost) b. We are using code to test in the meantime with User Streams, just to get an idea of what to do. I can see how after initial REST calls to prime the pump that Messages, Mentions, Timeline additions, Follows, Favorites, User Profile changes, etc will be streamed, and I can see how to deal with this. It is doable. However I do not understand how to keep track of some info in the Streams API, and wonder if these just will require the REST API 'forever': a:Consider a User who is not me (the authenticated user). I want to see their profile info. I do not see how the Site/User Stream will send that to me. b: Consider a User who is not being followed. I want to get their tweets (etc) via the streaming API. (I know how to get the tweets they sent via the REST api.) BUT: Since they are not being followed I will not have seen their tweets on the streaming API. Do I have to always use the REST API for this? If this is not the right place to ask these questions, please point me where to ask. thanks, Mark -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] Re: oauth_callback
Thanks Tim and Dave. It was not very clear in the docs, but I looked at the latest docs from the library I use (from jmathai) and I did find that he documented how to use this. I am up and running now from multiple domains. thanks, Mark On Dec 8, 4:07 am, Dave-twiends wrote: > Hi Tim, I'm pretty sure the oauth_verifier is documented in their > oAuth articles.. I'm speeking from memory here, but I'm sure I saw > last week when we were investigating our own oAuth issues.. > > But, nonetheless, you are correct, oauth_verifier should be passed > back every time. > > Dave > Twiends > > On Dec 8, 2:27 am, Tim Bull wrote: > > > There is a required OAuth parameter step which is unclearly documented > > by Twitter. When Twitter returns from your /oauth/authorize It returns > > an oauth_verifier token. Make sure that you pass this oauth_verifier > > token (along with the other parameters) along to > > your /oauth/access_token call. > > > Make sure you are passing this oauth_verifier in and see how you go. > > I've found that if you DON'T set a callback, it doesn't enforce the > > verifier, but if you do, then the verifier is essential (just be aware > > Twitter are planning to change to always require this in the future, so > > it's more compliant with the spec; worth making this change regardless, > > a lot of Twitter libraries don't implement it). > > > Hope this helps... > > > Tim -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
[twitter-dev] oauth_callback
I've been using a development site to work on a twitter related project until a few weeks ago, but I cut over to our real production site. Let's call development: mydev.com, and production: myprod.com. I had changed my twitter domain for callbacks to prod.com, no problem when I did this. Since I use cookies to manage which subdomain will be authenticated, I cannot have both dev and prod use cookies to get back to the correct domain (one of the cookies is the real url of the caller, so the dev domain cannot even see the cookie). I read about oauth_callback today, just to see if I could make this work for any general domain I might run my application on. First, I added mydev.com as a domain to my twitter app. Then I changed the requesttoken to use a simple oauth_callback on the dev site of http://mydev.com/proj/confirm (the registered callback is http://myprod.com/proj/confirm). When I call twitter on return -- and it returns to the right place!!! -, it gets an 'unexpected twitter failure' with nothing in any error string while calling my library's function to get the oauth token and secret. My question, finally: Did I miss some step here? Otherwise the code for mydev and myprod is identical. Do different domains have different ouath tokens in some way for the same application or something? thanks, Mark -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk
Re: [twitter-dev] Digest for twitter-development-talk@googlegroups.com - 25 Messages in 13 Topics
Kalucki Sep 09 08:02AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>Which URL are you requesting? What is the text message returned? Does
>this
>doc help? http://dev.twitter.com/pages/streaming_api_response_codes
>
>-John
>
>
>
>
>
> Topic: Latitudes and Longitudes in the "Location"
> Field<http://groups.google.com/group/twitter-development-talk/t/c10c90a980451137>
>
>brentos Sep 09 07:17AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>Hi Folks,
>
>Does anyone know from where these come? I'm speaking of the lat/lons
>in the actual "Location" field of the user, not the new per-tweet
>location stuff. It appears to be fixed for the vast majority of users
>to a single latitude and longitude. Do certain clients insert this
>once and then leave it?
>
>Thanks in advance for your answers!
>
>- Brent
>
>
>
>
>Taylor Singletary Sep 09 07:34AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>These are generally saved once by mobile-based Twitter clients, as
>you've
>surmised!
>
>Taylor
>
>
>
>
> Topic: Friends and Followers Resources for Protected
> Accounts<http://groups.google.com/group/twitter-development-talk/t/7c8cd3bfe7648a0f>
>
>Vega Sep 09 07:20AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>Going through the website. I was able to jump from a list to a users
>page. Even though the user is protected I can see their followers and
>who they are following. If I use the api against that user to get the
>users follower ids I get:
>"The remote server returned an error: (401) Unauthorized." Here is the
>documentation for the endpoint I am using,
>http://dev.twitter.com/doc/get/followers/ids.
>
>Seems like an inconsistency between the website and the api. I just
>want to verify that is excepted result from the api.
>
>Thanks for any help ahead of time.
>
>
>
> Topic: List Widget code upgraded
> yet?<http://groups.google.com/group/twitter-development-talk/t/ecd42a17fcabe402>
>
>Mike Sep 09 06:45AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>Hi Matt, this is the code.
>
>
>
>http://widgets.twimg.com/j/2/widget.js"</a>;>
>
>new TWTR.Widget({
>version: 2,
>type: 'list',
>rpp: 16,
>interval: 8000,
>title: 'Foo whatever',
>subject: '',
>width: 'auto',
>height: 300,
>theme: {
>shell: {
>background: '#FF',
>color: '#FF'
>},
>tweets: {
>background: '#FF',
>color: '#313131',
>links: '#2CA4E3'
>}
>},
>features: {
>scrollbar: false,
>loop: true,
>live: true,
>hashtags: true,
>timestamp: true,
>avatars: true,
>behavior: 'default'
>}
>}).render().setList('cnn', 'cnnnews').start();
>
>
>
>
>Thanks, Mike
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Topic: from:me in
> search<http://groups.google.com/group/twitter-development-talk/t/8e286d8413f21b34>
>
>Mark Krieger Sep 09 06:42AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>If I search for me by doing 'from:markskrieger' in either twitter
>directly, or from a twitter saved search of from:markskrieger, or from
>the equivalent in the api, I do not get recent results. In fact, I am
>not sure I am getting all results at all, it seems I am getting one or
>two tweets, others are just missing.
>
>Is this a known problem?
>
>Mark
>
>
>
>
>Taylor Singletary Sep 09 07:07AM -0700
> ^<#12af728c13d2126a_digest_top>
>
>All tweets don't appear in search, unfortunately. This article explains
>a
>bit about it:
>
>
> http://support.twitter.com/articles/66018-my-tweets-or-hashtags-are-missing-from-search
>
>Taylor
>
>
>
>
> --
> Twitter developer documentation and resources: http://dev.twitter.com/doc
> API updates via Twitter: http://twitter.com/twitterapi
> Issues/Enhancements Tracker:
> http://code.google.com/p/twitter-api/issues/list
> Change your membership to this group:
> http://groups.google.com/group/twitter-development-talk?hl=en
>
--
Mark Krieger
markskrie...@gmail.com
732-715-3074
Skype: markskrieger
Note: I am now using gmail.com, please update your email entry for me.
Thanks.
--
Twitter developer documentation and resources: http://dev.twitter.com/doc
API updates via Twitter: http://twitter.com/twitterapi
Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list
Change your membership to this group:
http://groups.google.com/group/twitter-development-talk?hl=en
[twitter-dev] from:me in search
If I search for me by doing 'from:markskrieger' in either twitter directly, or from a twitter saved search of from:markskrieger, or from the equivalent in the api, I do not get recent results. In fact, I am not sure I am getting all results at all, it seems I am getting one or two tweets, others are just missing. Is this a known problem? Mark -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk?hl=en
[twitter-dev] Re: Dynamic URLs with oauth return url
Jeff, All of my 'sites' are under control of one domain, so I just set a cookie before auth on any of the sites (with it's url), and then I redirect to the correct subdomain from my main site once my main site gets back control (I also do some housekeeping). I said 'I just set...' but this was fairly complex to get right and to deal with all situations. If you are going to many other sites, not under your control, I am not sure if the cookie method would work, I hope someone else has some ideas :) Mark On Sep 7, 1:37 am, Jeff Gladnick wrote: > I work forhttp://www.greatdentalwebsites.comand am trying to > configure the twitter integration to work with the new oauth system. > > The problem is that our users, when granting access, need to be > redirected back to their own website. The process works like this > > 1) Dentist is onhttp://theirdentalwebsite.comand they click the link > to "let my dental website talk to twitter" > 2) They are forwarded to the twitter page, and they click yes > 3) They are directed back to our website,http://www.greatdentalwebsites.com > to a special return url > > The problem is when they get to #3, I don't know how to determine > which customer's website to send them back to after that. All I have > in the url is their auth token thingy. > > I can think of two ways to solve this: > > 1) Pass along an additional url back from twitter somehow so when they > hit our return url, we have their domain name as a url parameter > 2) Pass the user back to their site in the first place, and handle the > return from twitter logic there. > > Has anyone else had to deal with similar problems and how was it > solved? -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk?hl=en
[twitter-dev] Re: More button
Thanks for the suggestion. I think I probably have a good but not perfect solution which is much less work and also not likely to ever fail: 1 - favorites and specific user updates have a count associated with them at any time. I can use that count to know when I am done. Pretty much. 2 - friends/followers/lists all have a next_cursor, so does a search, that works for me now. 3- Home timeline is an issue, but I can see how twitter will find that difficult to count for me or give me a next_cursor, etc. I guess if someone clicks 'more' once in a long while and I have to refresh to the same page with no more button because my one extra read showed no more updates, then I can live with that. Unless someone has a better idea? Mark On Aug 11, 11:33 am, Tom van der Woerdt wrote: > You could simply request twice as many as you need and do the math. > However, in rare occasions (very rare) it could happen that an user sent > 20 tweets and deleted all of them, in which case it may look like you > are at the end of the list. > > It is not recommended to use two API calls for 20 messages. > > Tom > > On 8/11/10 5:30 PM, Mark Krieger wrote: > > > Thanks for the quick response, that is what I thought I had > > remembered. > > > Does this mean that I always need to read-ahead to see if I am on the > > last page? > > > Mark > > > On Aug 11, 10:48 am, Julio Biason wrote: > >> On Wed, Aug 11, 2010 at 8:43 AM, Mark Krieger > >> wrote: > >>> 2. Assume that these calls will always really send me back 20 results > >>> ALWAYS, then if less, I know this is the last screen. > > >> Unfortunately, that won't work. Twitter retrieves the messages from > >> the cache and then tests if they still exist. This means you could > >> receive less than 20 and still not being in the last page. > > >> -- > >> Julio Biason > >> Twitter:http://twitter.com/juliobiason
[twitter-dev] Re: More button
Thanks for the quick response, that is what I thought I had remembered. Does this mean that I always need to read-ahead to see if I am on the last page? Mark On Aug 11, 10:48 am, Julio Biason wrote: > On Wed, Aug 11, 2010 at 8:43 AM, Mark Krieger wrote: > > 2. Assume that these calls will always really send me back 20 results > > ALWAYS, then if less, I know this is the last screen. > > Unfortunately, that won't work. Twitter retrieves the messages from > the cache and then tests if they still exist. This means you could > receive less than 20 and still not being in the last page. > > -- > Julio Biason > Twitter:http://twitter.com/juliobiason
[twitter-dev] More button
My app shows a list of updates from a call like UserTimeline or Favorites, gotten 20 at a time. It shows a More button at the bottom of the screen. Of course I do not want a More button on the last screen, I can think of 2 possible ways to do this: 1. Do 2 calls (argh) each time, see if the second call gives back no results, in that case I know that no More is needed. 2. Assume that these calls will always really send me back 20 results ALWAYS, then if less, I know this is the last screen. I see nothing in the return which shows that there is no next list to get. I am very wary of my #2 above. Does this mean I need to make 2 calls instead of one to know I am at the end? thanks, as usual, Mark
[twitter-dev] Re: Some twitterapi updates unavailable
Taylor,
I found one other in my debug logs: 20486403894, also from twitterapi.
None others yet. Hope this helps you.
Mark
On Aug 9, 9:29 am, Taylor Singletary
wrote:
> Hi Mark,
>
> We're looking into this and are not quite sure what's going on with these
> particular statuses. If you come across any other status ids that can't be
> fetched via statuses/show, cannot be favorited, or retweeted (all three
> actions fail with these particular tweets), please let us know the status id
> so we're aware.
>
> Thanks,
> Taylor
>
> On Mon, Aug 9, 2010 at 3:48 AM, Mark Krieger wrote:
> > I tried to 'favorite' an update by twitterapi over the weekend in
> > twitter, nothing happened, so I tried to read that update in my own
> > application -- and I then tried to read a few other updates from
> > twitterapi. I get back a statuses list in home timeline, then I try to
> > read more information about the updates in question, and I always get
> > an error. I think twitter must also get an error, since those updates
> > seem to be unavailable there too. Here is what I see in my debug
> > output for one of these updates, any ideas?
>
> > {"request":"/1/statuses/show/20499308096.json","error":"No status
> > found with that ID."}
>
> > I get the ID from the updates listed in a getstatusesHometimeline. I
> > would suspect my application, but it does this for everything else
> > fine, and only a subset of updates/statuses from twitterapi seem to be
> > failing, and twitter itself is failing to allow a favorite (for
> > instance) on these. Are these just 'lost' statuses/updates which I
> > should not worry about?
>
> > thanks,
>
> > Mark
[twitter-dev] Some twitterapi updates unavailable
I tried to 'favorite' an update by twitterapi over the weekend in
twitter, nothing happened, so I tried to read that update in my own
application -- and I then tried to read a few other updates from
twitterapi. I get back a statuses list in home timeline, then I try to
read more information about the updates in question, and I always get
an error. I think twitter must also get an error, since those updates
seem to be unavailable there too. Here is what I see in my debug
output for one of these updates, any ideas?
{"request":"/1/statuses/show/20499308096.json","error":"No status
found with that ID."}
I get the ID from the updates listed in a getstatusesHometimeline. I
would suspect my application, but it does this for everything else
fine, and only a subset of updates/statuses from twitterapi seem to be
failing, and twitter itself is failing to allow a favorite (for
instance) on these. Are these just 'lost' statuses/updates which I
should not worry about?
thanks,
Mark
[twitter-dev] Re: DMs and followers list
Thank you, exactly what I wanted to know, Mark On Aug 6, 1:22 pm, Matt Harris wrote: > Hi Mark, > > In the Twitter direct message composition screen we display an alphabetised > list of up to 100 of the most recent users you sent direct messages to. > > Hope that helps, > Matt > > On Wed, Aug 4, 2010 at 12:38 PM, Mark Krieger wrote: > > > > > Thanks, one of my colleagues, who is more ajax-savvy than me, > > suggested similar just a few > > minutes ago. > > > I appreciate your advice. We will do this, but I probably not top > > priority. > > > I still wonder what twitter does, since someone I know who does have > > many followers could not > > explain what twitter does with his DM followers dropdown > > > Thanks again, > > > Mark > > > On Aug 4, 11:40 am, Marcelo Calbucci wrote: > > > Mark, you should implement similarly to what Google Auto-suggest does... > > As > > > users start typing they send AJAX calls back to the servers that return > > the > > > top 10 matches. As long as your server responds quickly, users are not > > going > > > to notice because the total latency might be 100 or 200ms, which would > > > enough to keep users happy. > > > > -Marcelo > > > > On Wed, Aug 4, 2010 at 6:32 AM, Mark Krieger > > wrote: > > > > I am working on a twitter application, I want followers to be kept in > > > > a dropdown in my application like twitter does when I send a DM from > > > > twitter. However, since I have less than 50 followers now (alas), I do > > > > not know the twitter behavior when someone has (let's say) 5000 or > > > > 1 or even 10 followers. Surely you do not put a gigantic list > > > > into that dropdown. But since I cannot test that situation, can > > > > someone tell me what twitter itself does? I'd like to more or less do > > > > the same as twitter in this case. > > > > > thanks, > > > > > Mark > > > > -- > > > -Marcelo > > > > Twitter: @calbucci <http://twitter.com/calbucci> | blog.calbucci.com | > > Seattle > > > 2.0 <http://www.seattle20.com/> > > -- > > Matt Harris > Developer Advocate, Twitterhttp://twitter.com/themattharris
[twitter-dev] Re: DMs and followers list
Thanks, one of my colleagues, who is more ajax-savvy than me, suggested similar just a few minutes ago. I appreciate your advice. We will do this, but I probably not top priority. I still wonder what twitter does, since someone I know who does have many followers could not explain what twitter does with his DM followers dropdown Thanks again, Mark On Aug 4, 11:40 am, Marcelo Calbucci wrote: > Mark, you should implement similarly to what Google Auto-suggest does... As > users start typing they send AJAX calls back to the servers that return the > top 10 matches. As long as your server responds quickly, users are not going > to notice because the total latency might be 100 or 200ms, which would > enough to keep users happy. > > -Marcelo > > On Wed, Aug 4, 2010 at 6:32 AM, Mark Krieger wrote: > > I am working on a twitter application, I want followers to be kept in > > a dropdown in my application like twitter does when I send a DM from > > twitter. However, since I have less than 50 followers now (alas), I do > > not know the twitter behavior when someone has (let's say) 5000 or > > 1 or even 10 followers. Surely you do not put a gigantic list > > into that dropdown. But since I cannot test that situation, can > > someone tell me what twitter itself does? I'd like to more or less do > > the same as twitter in this case. > > > thanks, > > > Mark > > -- > -Marcelo > > Twitter: @calbucci <http://twitter.com/calbucci> | blog.calbucci.com | Seattle > 2.0 <http://www.seattle20.com/>
[twitter-dev] DMs and followers list
I am working on a twitter application, I want followers to be kept in a dropdown in my application like twitter does when I send a DM from twitter. However, since I have less than 50 followers now (alas), I do not know the twitter behavior when someone has (let's say) 5000 or 1 or even 10 followers. Surely you do not put a gigantic list into that dropdown. But since I cannot test that situation, can someone tell me what twitter itself does? I'd like to more or less do the same as twitter in this case. thanks, Mark
[twitter-dev] Re: Undocumented fields?
Wonderful, thanks. VERY helpful. Exactly the problem we were having. Mark On Jul 23, 1:52 pm, Matt Harris wrote: > Hi Mark, > > These fields were added in January 2010 to address the problem when the > cursors were too long for javascript to handle. The original announcement is > on our API Announce list [1] and there are no plans to remove this > functionality. > > Hope that helps, > Matt > > 1.http://groups.google.com/group/twitter-api-announce/browse_thread/thr... > > On Fri, Jul 23, 2010 at 8:45 AM, Mark Krieger wrote: > > > I found that statuses/friends and statuses/followers have an > > undocumented field called next_cursor_str and prev_cursor_str, which > > are simply the string values of those fields in addition to the digits > > passed back in next_cursor and prev_cursor. Is this planned to stay > > this way? (can I rely on them always being in the api?) > > > I was working around a known json_decode bug on 32-bit machines on one > > of our test portables, where it could not decode the next_cursor, no > > big deal, but when I found the 'str' versions of the cursors I turned > > off the workaround. Once $id's get large enough, 32-bit machines will > > need json_decode to work right tooso the workaround might be > > needed then. > > > Mark > > -- > > Matt Harris > Developer Advocate, Twitterhttp://twitter.com/themattharris
[twitter-dev] Undocumented fields?
I found that statuses/friends and statuses/followers have an undocumented field called next_cursor_str and prev_cursor_str, which are simply the string values of those fields in addition to the digits passed back in next_cursor and prev_cursor. Is this planned to stay this way? (can I rely on them always being in the api?) I was working around a known json_decode bug on 32-bit machines on one of our test portables, where it could not decode the next_cursor, no big deal, but when I found the 'str' versions of the cursors I turned off the workaround. Once $id's get large enough, 32-bit machines will need json_decode to work right tooso the workaround might be needed then. Mark
[twitter-dev] Re: getStatusesMentions not working today....???
Matt,
I found the problem, sorry, I believe it is on my end. I found a line
of code someone here changed. Argh. I will be more careful with your
time in the future.
Mark
On Jul 21, 6:00 pm, Mark Krieger wrote:
> Matt,
>
> Thanks for the quick response. The call uses EpiTwitter.php, it looks
> like:
>
> Call Oauth to authenticate (this works), then:
>
> $val = $twitterObj->get_statusesMentions(array('since_id' =>
> $lastid));
> $arr = json_decode($val->responseText, true);
> if (empty($arr)) { $error = "error text"; log_message(...)}
> else ...
>
> It logs the "error text" showing the response through EpiTwitter was
> empty. Also, when I print $val->responseText, it is EMPTY.
>
> I've tried it on 4 twitter accounts I have, all of which have worked
> for many months in code which has not changed, it has only failed
> before when the api was down (like tuesday). And all other api calls
> which use similar code all work fine. I am stumped.
>
> I can send more data, like the $lastid, and my credentials for oauth
> to you, I would of course do that privately. I will run some more
> tests from here first I guess.
>
> Thanks,
>
> Mark
> On Jul 21, 4:49 pm, Matt Harris wrote:
>
> > Hi Mark,
>
> > We aren't seeing any errors like this when we run some tests. Could you
> > elaborate on what the bogus headers are?
>
> > Matt
>
> > On Wed, Jul 21, 2010 at 12:51 PM, Mark Krieger
> > wrote:
>
> > > My application has used getStatusesMentions for many months, it is
> > > only failing when twitter has had some problem, like two days ago, and
> > > in that case, everything is flaky. Today it is failing, but everything
> > > else works fine. Is this a known problem today? getStatusesMentions
> > > always comes back today with bogus headers
>
> > > Mark
>
> > --
>
> > Matt Harris
> > Developer Advocate, Twitterhttp://twitter.com/themattharris
[twitter-dev] Re: getStatusesMentions not working today....???
Matt,
Thanks for the quick response. The call uses EpiTwitter.php, it looks
like:
Call Oauth to authenticate (this works), then:
$val = $twitterObj->get_statusesMentions(array('since_id' =>
$lastid));
$arr = json_decode($val->responseText, true);
if (empty($arr)) { $error = "error text"; log_message(...)}
else ...
It logs the "error text" showing the response through EpiTwitter was
empty. Also, when I print $val->responseText, it is EMPTY.
I've tried it on 4 twitter accounts I have, all of which have worked
for many months in code which has not changed, it has only failed
before when the api was down (like tuesday). And all other api calls
which use similar code all work fine. I am stumped.
I can send more data, like the $lastid, and my credentials for oauth
to you, I would of course do that privately. I will run some more
tests from here first I guess.
Thanks,
Mark
On Jul 21, 4:49 pm, Matt Harris wrote:
> Hi Mark,
>
> We aren't seeing any errors like this when we run some tests. Could you
> elaborate on what the bogus headers are?
>
> Matt
>
> On Wed, Jul 21, 2010 at 12:51 PM, Mark Krieger wrote:
>
> > My application has used getStatusesMentions for many months, it is
> > only failing when twitter has had some problem, like two days ago, and
> > in that case, everything is flaky. Today it is failing, but everything
> > else works fine. Is this a known problem today? getStatusesMentions
> > always comes back today with bogus headers
>
> > Mark
>
> --
>
> Matt Harris
> Developer Advocate, Twitterhttp://twitter.com/themattharris
[twitter-dev] getStatusesMentions not working today....???
My application has used getStatusesMentions for many months, it is only failing when twitter has had some problem, like two days ago, and in that case, everything is flaky. Today it is failing, but everything else works fine. Is this a known problem today? getStatusesMentions always comes back today with bogus headers Mark
[twitter-dev] Re: Statuses/friends previous cursor broken
Matt, Thank you. This has been open for about 4 months. I hope it will be fixed soon. For now I think I will need to do something on my own to program around it. Wish me good luck (: Mark On Jul 21, 1:37 pm, Matt Harris wrote: > Thanks for the emails on this one. It is a known issue which our engineering > team are looking into. It is being tracked on our issue tracker > here:http://code.google.com/p/twitter-api/issues/detail?id=1572 > > When we have news on the progress we'll update the ticket. > > Best, > Matt > > > > On Wed, Jul 21, 2010 at 8:32 AM, Scott Marshall wrote: > > Yup, I encountered this as well while trying to figure out count > > differences in what twitter reports on the number of followers and > > what the api gives. Seems to be off by one number. So flipping > > between pages on twitter itself (followers list) I did get those blank > > results. I found I had blocked a user a while back with twitter, and > > this person was still being listed with results in the api but not on > > twitter itself. Might have thrown off the count that way. Took care > > of the so called blocked user, but unblocking and reblocking it. > > count was still off. But the follower was no longer in my api > > results. Which is a good thing. Still not sure why counts differ. > > > anyways, sorry for bringing up this count difference. I just wanted > > to say that I've came across the "blank" pages issue while flipping > > back thru the followers list like Mark had mentioned. > > > On Wed, Jul 21, 2010 at 11:17 AM, Mark Krieger > > wrote: > > > Hi. I am having the same problem as several others reported, where the > > > result of a statuses/friends sends back a previous cursor which is not > > > correct. The thread on this, from april or may, died with no > > > resolution. I notice that twitter itself has this same problem, > > > although twitter seems to use page=nnn instead of cursor=nnn, if you > > > try to go back to previous, it will list no users. > > > > Is a fix planned for this? > > > > thanks, > > > > Mark > > -- > > Matt Harris > Developer Advocate, Twitterhttp://twitter.com/themattharris
[twitter-dev] Statuses/friends previous cursor broken
Hi. I am having the same problem as several others reported, where the result of a statuses/friends sends back a previous cursor which is not correct. The thread on this, from april or may, died with no resolution. I notice that twitter itself has this same problem, although twitter seems to use page=nnn instead of cursor=nnn, if you try to go back to previous, it will list no users. Is a fix planned for this? thanks, Mark
