Greasy Fork

The request frequency of this script is way too fast. Although the problem has been solved, I am seeing a lot of 400 errors in the console which is quite alarming. It also causes the number of running scripts displayed in my Tampermonkey to rapidly increase by several tens, even though there are actually only a few scripts. Is there any way to improve this issue?

Sup @dlzrncsb, the frequency does not actually matter because the method used doesn't actually send requests to OpenAI endpoint (it just refreshes an invisible iframe w/ latest SSG manifest to keep content fresh, without using an API so it won't be logged).

Re: 400 errors, those are intentionally desired responses that happen as a result of POST requests being sent by beacons (OpenAI only supports GET) just to keep Chrome tab from sleeping without having to set chrome://discards/ flag (so less user action required).

But since it concerns you, I will work on allowing you to customize frequency, and also choose a different endpoint than OpenAI

Re: running scripts number, this is happening from other scripts, they are re-injecting their code into the invisible iframe everytime it refreshes to keep session alive. If you really like using those scripts, you should recommend to the authors to add `// @noframes` to prevent their scripts from loading in frames (will improve compatibility with other scripts that use headless methods too)

@dlzrncsb if they are concerned with performance, tell them it will actually improve since it will make their script only run on the main frame that matters (so less resources consumed)

@dlzrncsb I updated the script to allow user input for refresh interval per your concern.

Also, requests when ChatGPT is background Chrome tab now send to httpbin.org instead of openai.com (but results in CORBS error because domain is no longer openai.com, but doesn't matter because the only purpose is to trick Chrome into thinking tab is active so it doesn't sleep)

Can you update please?

Thank you very much for your update and detailed explanation. I will investigate the issue of the rapidly increasing number of scripts more thoroughly. It may be caused by something happening between the scripts, or perhaps it is indeed my own issue

@dlzrncsb it's because when the iframe src is updated, the other script injects again if they don't specify `// @noframes` (which all ChatGPT scripts should do, to only focus on ChatGPT main window) but the rising number does not mean those scripts are running that many times, it just measured how many times it got injected. But every time it re-injects, the old version is already cleared away, so it's not as bad as it looks.

But if you suggest to these authors to use `// @noframes`, then their script won't cause worry with the visually large numbers, and it will perform even slightly better because it only injects once if other scripts use iframes (but again, old instances are cleared away so not that much better, but for peace of mind for user)