If I use the New Twitch API (which I always do, naturally) to check how long a given user has been streaming, it will report the time when they started the broadcast.
However, some of them don’t play the same game for all that time, but switch between games, sometimes many times. This means that when I make my check, the “start time” essentially “lies” in the sense that it only tells me when the person started streaming, and not when they last switched games.
Is there really no way to get this information? If not, please add a separate timestamp which corresponds to when they started streaming for the current game. (Assuming of course that they update this – some forget to do that, and then Twitch obviously can’t detect this on its own.)
Other statistics people don’t have a problem with this…
Make one!
Sounds you you seem to be collecting data for no particular user. Best make sure whatever you are doing abides by the Developer TOS that you already agreed to when you created a Twitch ClientID.
Considering further more you are hiding behind a fake name.
Specifically around data collection and storage. Additionally with GDPR when collecting data in the EU or for users from the EU
It’s not missing. It’s in the streams end point if you continually poll for data changes, including but not limited to, category, stream title, tags, or viewer count
I don’t know what “statistics people” are or why you think I’m one, but if anyone is okay with this, they must be deranged. Or maybe they just realize that it’s not worth it trying to nag you about it…
Seriously? I can’t even begin to tell you how impractical and stupid this is. You can’t be serious.
Why would I want to go through the hell of dealing with hosting companies, domain registrars, DNS hosts and all their BS, paying them a bunch of money all the time, just to use the Twitch API? I don’t get paid anything for doing so, and even if I did collect data (which I don’t), it would just be meaningless noise. Twitch is a bunch of kids screaming in chat rooms – not some sort of ultra-curated goldmine of personal information. Either way, nothing I’ve asked for would in any way be “worse for privacy” compared to the existing API.
I’m actually not collecting any data whatsoever. I’m trying to use the Twitch API, but it’s been (seemingly) deliberately crippled as to be practically useless.
I’m not “hiding” behind anything. This forum forced me to pick a nickname to post, so I had to enter something. That’s it. Why does it even matter who I am?
I already explained why this doesn’t make any sense, but considering the replies I’ve got so far while asking API questions, I feel as if you aren’t exactly eager to implement or fix anything. You seem to have very little interest in correcting even the most jarring bugs with the API, and the general tone of your messages make me think you’re basically mocking me…
I don’t have a whole infrastructure set up to poll your API 24/7 for every single user on the site. Even if that would be possible (it would obviously quickly hit the rate limits), it would add an insane amount of strain on your servers for no reason other than “we don’t wanna provide this basic info because you can do it in this extremely cumbersome and insane way instead”. But why? I seriously don’t understand your mentality. If I were running Twitch and somebody pointed out the things I’ve asked about (perhaps with the exception of the rate limit headers, which I did figure out how to deal with eventually), I would be happy that somebody identified them and probably fix them within hours – not talk about how it’s “not on the roadmap”.
This almost makes me wonder why you have an API at all. I feel as if I’m missing something.
Logically, if you didn’t care about feedback, you wouldn’t even have a forum like this. But since you do, why not actually listen to what people report? At least when it comes to simple things like the ones I’ve mentioned? And finally, sorry if you have something personally against me because I “sound rude” or something – I’m just very to-the-point, and when I feel like I’m not being told the truth, I get very frustrated.
Then you would lack a server of sufficient size, processing power, or web accessibility to do much, but that does depend on your use case, I’ve had no issue with any of these things you have mentioned for 15 years at least
If that’s what you think, why bother. Thank you for insulting everyone.
And the rate limit should suffice for long polling all active streams for this data, I can think of a few ways to optimize the calls to reduce API requests anyway, it’s all in the documentation.
If you are not working with single streamers and want to obtain game changes it’s in real time at
or if you prefer kraken
That’s how other people do it, just long poll all streams and collect the data.
I don’t not work for Twitch. I cannot make changes to the API.
I am a third party developer who uses the Twitch API, and advise on the forums how I do things when people raise questions.
I long poll to obtain accurate game change data for a stream. Or since WebHooks have been introduced, I now use webhooks for more accurate data collection.
Comparing API’s across various different system over the years, having Webhooks is a great addition to the Suite of tools available.
Again, I’m not a Twitch Employee, I don’t not work on or for Twitch on the API
It may seem simple to you, but that doesn’t mean it is.
As a moderator for the forums, yes, I have concerns when you use such a name.
I am telling the truth, if I want to obtain this data, I do exactly what, for example, sullygnome or other statistics people do, I poll all streamers that are live and test the returned data objects for game differences and record the time.
Could this be done a better way with a historical API data end point?