[twitter-dev] Re: API X-RateLimit-Remaining goes to 0 on our first request after midnight
This happened again at midnight of the 16th. The last request I made on the 15th was at 11:59:15 PM with a reset of 12:27:54 AM and a remaining of 144. The next request I made was at 12:19:15 AM with the same reset and a remaining of 0. The next successful request I made was at 12:29:15 AM. Is there a better place I should be posting this? Like some sort of bug tracking application rather than just a development talk? -Zach -- 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: API X-RateLimit-Remaining goes to 0 on our first request after midnight
Just got into the office and sure enough I got an email of a 400 response last night at 12:59 AM. From what I can tell the last request before the 400 was at 11:55 PM and had an expiration time of 11:57 PM. This does still match up with my first request after midnight theory, but I noticed something else interesting in the header. The date in the header is Sat, 15 Jan 2011 06:59:21 GMT, which would be 12:59:21 AM CST. The X-RateLimit-Reset is 1295074762, which is 12:59:22 AM CST. It's possible that there was a request done after 11:57 PM that I'm not tracking and that our request was just lucky to be sent one second before our reset, but I find the one second different a little too close to ignore. -Zach On Jan 14, 9:32 am, Zach Gardner z.gard...@hotmail.com wrote: We've been having problems recently with going over our rate limit when making API calls. We do a lot of server side caching of the responses we get from the API, so I was curious why we were going over our limit. I setup our application to record the headers we receive each time we make an API call, and found something interesting with X- RateLimit-Remaining: When we make our first API call in a while a little before midnight, we get a new rate limit remaining and new expiration time. But as soon as we make a call to the API after midnight, our rate limit goes to 0, so we have to wait till the expiration time to start making calls again. Yesterday at 11:50 PM CST we made the first call to the API after our expiration time passed. Our X-RateLimit-Remaining was 149 out of 150 because that call used one of our allotted 150 per hour. The next call we made to the API was this morning at 12:10 AM CST, but our X- RateLimit-Remaining was 0. It's entirely possible that I didn't add the header tracking to another place we do API calls, but I doubt we'd have that much traffic at that time of the day to use 149 calls in 20 minutes. For the past few weeks I had it set to email me when the API returns a 400 Bad Request header. I've gotten one on the 11th at 12:17, the 10th at 12:37, the 8th at 12:22, and the 6th at 12:23 AM CST. (Google crawls our site around midnight our time, and our server side caches have expired by that time, which is the origin of most of the requests.) I found something interesting while looking through our logs for the 12th. We made an API call on the 11th at 11:35 PM, which used our 146th call per hour. The expiration time for that hour was the 12th at 12:04 AM. The next call we made was on the 12th at 12:35 AM, after our expiration date. Our rate limit remaining was back at 149 out of 150 with an expiration at 1:35 AM as expected. I was wondering if anyone has experienced something similar, or am I just missing something entirely. I can provide the headers in full if necessary.. Thanks, -Zach -- 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] API X-RateLimit-Remaining goes to 0 on our first request after midnight
We've been having problems recently with going over our rate limit when making API calls. We do a lot of server side caching of the responses we get from the API, so I was curious why we were going over our limit. I setup our application to record the headers we receive each time we make an API call, and found something interesting with X- RateLimit-Remaining: When we make our first API call in a while a little before midnight, we get a new rate limit remaining and new expiration time. But as soon as we make a call to the API after midnight, our rate limit goes to 0, so we have to wait till the expiration time to start making calls again. Yesterday at 11:50 PM CST we made the first call to the API after our expiration time passed. Our X-RateLimit-Remaining was 149 out of 150 because that call used one of our allotted 150 per hour. The next call we made to the API was this morning at 12:10 AM CST, but our X- RateLimit-Remaining was 0. It's entirely possible that I didn't add the header tracking to another place we do API calls, but I doubt we'd have that much traffic at that time of the day to use 149 calls in 20 minutes. For the past few weeks I had it set to email me when the API returns a 400 Bad Request header. I've gotten one on the 11th at 12:17, the 10th at 12:37, the 8th at 12:22, and the 6th at 12:23 AM CST. (Google crawls our site around midnight our time, and our server side caches have expired by that time, which is the origin of most of the requests.) I found something interesting while looking through our logs for the 12th. We made an API call on the 11th at 11:35 PM, which used our 146th call per hour. The expiration time for that hour was the 12th at 12:04 AM. The next call we made was on the 12th at 12:35 AM, after our expiration date. Our rate limit remaining was back at 149 out of 150 with an expiration at 1:35 AM as expected. I was wondering if anyone has experienced something similar, or am I just missing something entirely. I can provide the headers in full if necessary.. Thanks, -Zach -- 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: Mad about lists and cursors... please help
I don't know if this fix for next_cursor always being zero has been deployed or not, but I'm still seeing this bug. A fix for this would be really awesome. On Apr 17, 12:04 pm, Mark McBride mmcbr...@twitter.com wrote: Yes. A fix has been identified, and should be deployed in a few days Sent from mobile device On Apr 17, 2010, at 7:08 AM, Zach zcox...@gmail.com wrote: It's 10 days later and next_cursor on http://apiwiki.twitter.com/Twitter-REST-API-Method%3A-GET-list-member... is still always 0, even when the user is being followed by far more than 20 lists. This is completely broken and prevents 3rd party apps from discovering all lists that follow a given user. Has anyone at Twitter even looked into this? On Apr 7, 3:43 pm, Mark McBride mmcbr...@twitter.com wrote: Eugene, we're aware of the issue and will take a look at it today. ---Mark http://twitter.com/mccv On Wed, Apr 7, 2010 at 11:09 AM, eugene.man...@gmail.com eugene.man...@gmail.com wrote: I posted this issue to @twitterapi twice, but they ignored it. Dear API group, please address this question. Thank you! On Apr 6, 9:45 am, Spraycode joey.fernan...@gmail.com wrote: Has anyone been able to solve this issue? This is still crippling us. Thanks! On Apr 2, 5:25 am, luisfigo rsoeg...@gmail.com wrote: Having the same problem... Triedhttp://api.twitter.com/1/avinashkaushik/lists/memberships.xml and get 0 forcursor. This guy is followed by ton of lists in fact Below is the snapshot of the end result I got... This is screwing up our app right now... . profile_background_image_url http://a1.twimg.com/profile_background_images/79104366/twitter_backgr ... /profile_background_image_url profile_background_tilefalse/profile_background_tile notificationsfalse/notifications geo_enabledfalse/geo_enabled verifiedfalse/verified followingfalse/following statuses_count3208/statuses_count langen/lang contributors_enabledfalse/contributors_enabled /user /list /lists next_cursor0/next_cursor previous_cursor0/previous_cursor /lists_list On Apr 1, 6:00 pm, Diego Rin Martin diego@gmail.com wrote: I think it's a API bug, even in the twitter page the paginator doesn't work as expected, sometimes appears, sometines not, and when appears it makes in a random manner. i'm gettingcursor0 from API, using int or string representation, the bug is in the API that sends thecursor0 randomly. regards, diego. On Thu, Apr 1, 2010 at 2:38 AM, jmathai jmat...@gmail.com wrote: Are you sure you're using the string representation of the cursor instead of the int? The API'scursorexceeds PHP's max integer value (generally). jmathai ~ $ php -r '$x = json_decode (1); echo $x; echo \n; var_dump ($x===1); var_dump($x===1.111E+52);' 1.111E+52 bool(false) bool(true) jmathai ~ $ php -r '$x = 1; echo $x; echo \n; var_dump ($x===1); var_dump($x===1.111E+52);' 1.111E+52 bool(true) bool(false) On Mar 31, 2:03 am, Diego Rin Martín diego@gmail.co m wrote: Hi there, this is my first post to this group, i'm a spanish developer dealing with twitter api surprises, excuse my poor english, i'will do my best to comunicate nicest. So, to the problem, I'm trying to retrieve the lists for a user, via list/membershipsget method, and passingcursoras parameter, I'm having got random results, I explain myself, sometimes I made a request (for user edans, that have a huge amount of pages to paginate) and I get one page, I passcursor-1 and I getcursor0, sometimes I get one page, I passcursor-1 i getcursor 1331431515904087602, then I pass it and I get 0, sometimes I get a random number of pages, but never, never, be able to retrieve the total amount of pages. I use php twitter-async classes to comunicate with API, I thought that it could be the cause of the problem, but using direct curl (via php5- curl extension) calls I'm having the same issues. Same using json or xml. I'm always getting 200 responses, so the call finish in a correct way. any clue? I'm turning mad. Thanks in advance. diego. -- To unsubscribe, reply using remove me as the subject.
[twitter-dev] Re: GET list memberships paging is broken?
I'm also always seeing next_cursor=0 both using twitter4j and curl. On Apr 17, 7:26 am, Dan Checkoway dchecko...@gmail.com wrote: When using the GET list memberships API (http://api.twitter.com/1/twitterapi/lists/memberships.*), it looks like paging is broken. If the user is a member of 20 lists, you can never see anything beyond the first 20. I'm passing cursor=-1 (well, twitter4j is) on the first request, and I get back the first page of 20 lists, which is fine...but no matter what, I always get back: next_cursor:0, previous_cursor:0, next_cursor_str:0, previous_cursor_str:0 ...which prevents any paging beyond that first page of 20. This is the case no matter which user I've tried. What seems coincidental is that even on the twitter web site proper, only the first page of 20 is presented as well, with no way to page beyond that...for example: http://twitter.com/GamePro/lists/memberships I'm not sure if that's a related issue, or an intentional thing that has also affected the API, or what. Anyway, can twitter please fix paging on the GET list memberships API? Thanks, Dan Checkoway -- Subscription settings:http://groups.google.com/group/twitter-development-talk/subscribe?hl=en
[twitter-dev] Re: Mad about lists and cursors... please help
It's 10 days later and next_cursor on http://apiwiki.twitter.com/Twitter-REST-API-Method%3A-GET-list-memberships is still always 0, even when the user is being followed by far more than 20 lists. This is completely broken and prevents 3rd party apps from discovering all lists that follow a given user. Has anyone at Twitter even looked into this? On Apr 7, 3:43 pm, Mark McBride mmcbr...@twitter.com wrote: Eugene, we're aware of the issue and will take a look at it today. ---Mark http://twitter.com/mccv On Wed, Apr 7, 2010 at 11:09 AM, eugene.man...@gmail.com eugene.man...@gmail.com wrote: I posted this issue to @twitterapi twice, but they ignored it. Dear API group, please address this question. Thank you! On Apr 6, 9:45 am, Spraycode joey.fernan...@gmail.com wrote: Has anyone been able to solve this issue? This is still crippling us. Thanks! On Apr 2, 5:25 am, luisfigo rsoeg...@gmail.com wrote: Having the same problem... Triedhttp://api.twitter.com/1/avinashkaushik/lists/memberships.xml and get 0 for cursor. This guy is followed by ton of lists in fact Below is the snapshot of the end result I got... This is screwing up our app right now... . profile_background_image_url http://a1.twimg.com/profile_background_images/79104366/twitter_backgr... /profile_background_image_url profile_background_tilefalse/profile_background_tile notificationsfalse/notifications geo_enabledfalse/geo_enabled verifiedfalse/verified followingfalse/following statuses_count3208/statuses_count langen/lang contributors_enabledfalse/contributors_enabled /user /list /lists next_cursor0/next_cursor previous_cursor0/previous_cursor /lists_list On Apr 1, 6:00 pm, Diego Rin Martin diego@gmail.com wrote: I think it's a API bug, even in the twitter page the paginator doesn't work as expected, sometimes appears, sometines not, and when appears it makes in a random manner. i'm getting cursor 0 from API, using int or string representation, the bug is in the API that sends the cursor 0 randomly. regards, diego. On Thu, Apr 1, 2010 at 2:38 AM, jmathai jmat...@gmail.com wrote: Are you sure you're using the string representation of the cursor instead of the int? The API's cursor exceeds PHP's max integer value (generally). jmathai ~ $ php -r '$x = json_decode(1); echo $x; echo \n; var_dump($x===1); var_dump($x===1.111E+52);' 1.111E+52 bool(false) bool(true) jmathai ~ $ php -r '$x = 1; echo $x; echo \n; var_dump($x===1); var_dump($x===1.111E+52);' 1.111E+52 bool(true) bool(false) On Mar 31, 2:03 am, Diego Rin Martín diego@gmail.com wrote: Hi there, this is my first post to this group, i'm a spanish developer dealing with twitter api surprises, excuse my poor english, i'will do my best to comunicate nicest. So, to the problem, I'm trying to retrieve the lists for a user, via list/membershipsget method, and passing cursor as parameter, I'm having got random results, I explain myself, sometimes I made a request (for user edans, that have a huge amount of pages to paginate) and I get one page, I pass cursor -1 and I get cursor 0, sometimes I get one page, I pass cursor -1 i get cursor 1331431515904087602, then I pass it and I get 0, sometimes I get a random number of pages, but never, never, be able to retrieve the total amount of pages. I use php twitter-async classes to comunicate with API, I thought that it could be the cause of the problem, but using direct curl (via php5- curl extension) calls I'm having the same issues. Same using json or xml. I'm always getting 200 responses, so the call finish in a correct way. any clue? I'm turning mad. Thanks in advance. diego. -- To unsubscribe, reply using remove me as the subject.
[twitter-dev] Re: oauth/authenticate error/bug
Yeah, there is a major issue with both oauth/authenticate and oauth/ authorize. Using oauth/authorize, a 403 forbidden is produced if 1) user logs in via my app and then tries to log into twitter.com or 2) user logs into twitter.com and then tries to log in via my app I believe that twitter.com's _twitter_sess cookie is doing something to produce a 403 whenever we try to authorize or authenticate an already signed in user. On Aug 12, 3:03 pm, aschobel ascho...@gmail.com wrote: I'm also getting this error when trying to block folks using twitter.com, so it may not be specific to oauth. On Aug 12, 2:53 pm, aschobel ascho...@gmail.com wrote: We are having the same issue, getting a 403 forbidden. I tried another OAuth enabled site and they have the same issue, http://www.twitlonger.com/ Maybe there is some type of outage? On Aug 12, 1:45 pm, Zach zwe...@gmail.com wrote: Use case: User logs in via oauth/authenticate User logs out via accounts/end_session. Cookies on my site containing access tokens are cleared. Same user logs back in later (in the same browser session) with oauth_authenticate. However, this last step produces a 403 forbidden: server understood request but refuses to fulfill it. Any ideas? This should be a fairly basic use case...
[twitter-dev] Re: OAuth and native clients
I'm going to 3rd Sebastian's POV. This is a serious problem that needs to be addressed now. Mobile app developers want to integrate their native apps with sites like Twitter and Facebook, but the current user experience is so unacceptable that no one is going to use it in its current form. For more on the topic: http://www.hueniverse.com/hueniverse/2009/02/beyond-the-oauth-web-redirection-flow.html Kudos to the Twitter team for actually starting to think about a reasonable solution to this problem. Facebook has the Connect for iPhone library, but even that is just their terrible JavaScript-based Connect login shown in an embedded browser. And forget about trying to authenticate with Facebook from something like a BlackBerry app. We are anxiously awaiting a OAuth for Mobile option for the Twitter API (as are a lot of other developers). Our app missed the from [MyApplication] using Basic Auth cutoff so now every status update we push says from Twitter4J... not the best for marketing purposes. We would also love not having to store passwords on the device and send them over the wire every time a user clicks the Share button. On Jun 30, 4:42 pm, morefromalan sbal...@gmail.com wrote: Just wanted to second Sebastian's POV here. UserExperience is a key revenue driver for us, andOAuthfor nativemobileapps is really painful for the user. On Jun 19, 5:41 am, Sebastian sdelm...@gmail.com wrote: Thanks for the pointer... I did some searches, but they were all focused onmobileclients. In my case, I'm not worried about the complexity of implementing OAuth. I can deal with that, and once it's done, it's gone from the picture. It's the user experience that worries me, as exposed on that thread by the TTYtter example. Well, since people are asking, the workflow doesn't significantly differ from otherOAuthapplications and depends on the fact that access tokens don't expire. When people start TTYtter up for the first time without an access token (or TTYtter tries the access token and it fails), it asks for the usual request token, prints the access URL with the request token it wants the user to authorize, and waits for the user to authorize. Twitter, presumably, will say, ok, tell your program to continue. Back on TTYtter's side, the user hits ENTER, and TTYtter exchanges its request token for an access token *and caches it* once it has verified it can successfully hit the user timeline for data. So far, this is not significantly different than any otherOAuthapp. Is there any other way to doOAuthand at the same time, behave like a sensible application? Could Twitter implement a basic auth api call to perform theoauth authorization in the first place? Such a call would only be allowed from clients that prove they need it, and could be revoked for rogue clients. I know this lowers the security ofOAuth, but it only officializes a hack many apps will try to implement. On Jun 19, 12:39 am, Cameron Kaiser spec...@floodgap.com wrote: Or is the door for basic auth really closing forever? This has been discussed in a number of threads and an exact determination has not yet been made. However, this might give you some context: http://groups.google.com/group/twitter-development-talk/browse_thread... -- personal:http://www.cameronkaiser.com/-- Cameron Kaiser * Floodgap Systems *www.floodgap.com*ckai...@floodgap.com -- The cost of living has not adversely affected its popularity. --- Hide quoted text - - Show quoted text -