Freezing & overlay issue

[walseb]

Hello!

I'm still having the issue shown in the screenshot in the screenshot on this issue #31, that is the issue where the overlay freezes forever, and executing the prompt again leads to the overlay ending up in the assistant block header (https://user-images.githubusercontent.com/2865774/239750507-4b96a63c-5633-4563-81b5-c17252d96062.png)

I'm running the latest commit of this repo. Any thoughts of how to find out what's happening when I get the issue?

[tyler-dodge]

I added additional logging with the latest change if you set org-assistant-logging to t.

That'll make it output to a buffer called org-assistant-logging. That might help make it easier to identify what is happening when it fails if this is still happening.

[walseb]

Excellent, thank you!

I will try it out and report my findings.

[walseb]

Thanks to your logging feature I was able to capture this when the error occurred. If I had to guess it's due to the overload error that OpenAI sends not being handled properly.

Deleted ("/tmp/jsoniUqvdk")
Response Complete Success: ("{
  \"error\": {
    \"message\": \"That model is currently overloaded with other requests. You can retry your request, or contact us through our help center at help.openai.com if the error persists. (Please include the request ID d478976008c1a18fff694b44c21a2b5d in your message.)\",
    \"type\": \"server_error\",
    \"param\": null,
    \"code\": null
  }
}
")
Request Body: (("--data" "@/tmp/jsonoZI4wx"))
Request Body: (("--data" "@/tmp/jsonBSnn62"))
Response: (((id . "chatcmpl-7QMZTpwsvxf5bCCDGRWxQZu3FDwaT") (object . "chat.completion.chunk") (created . 1686517647) (model . "gpt-3.5-turbo-0301") (choices . [((delta (role . "assistant")) (index . 0) (finish_reason))])))
Response: (((id . "chatcmpl-7QMZTpwsvxf5bCCDGRWxQZu3FDwaT") (object . "chat.completion.chunk") (created . 1686517647) (model . "gpt-3.5-turbo-0301") (choices . [((delta (content . "Chapter")) (index . 0) (finish_reason))])))
Deleted ("/tmp/jsonBSnn62")
Unexpected status ("killed
")
Deleted ("/tmp/jsonoZI4wx")
Response Complete Success: ("{
  \"error\": {
    \"message\": \"That model is currently overloaded with other requests. You can retry your request, or contact us through our help center at help.openai.com if the error persists. (Please include the request ID 574324db019caf11e314741760e6ecbd in your message.)\",
    \"type\": \"server_error\",
    \"param\": null,
    \"code\": null
  }
}
")

Thoughts?

[tyler-dodge]

Thanks for that info. I think that's enough for me to write a test around it. I'll probably have the time to fix it within the next week.