After the latest update, all of my notebooks are being read as ‘long files’ which wasn’t happening previously. Then, the generation never completes in the chat. I’ve tried a few models. Seem like others are having issues with core functionality, would appreciate an ability to downgrade.
I am unable to reproduce on my end.
Could you screen record this behavior and send to aman@cursor.com? Let me know here when you’ve sent it.
Apologies - I think the new ui tripped me up and I was incorrect about this behavior.
Have you changed how the output cells of jupyter notebooks are handled though? I think the long file thing popping up is probably a result of inline images being inserted into the context, so notebooks i previously had no issues with are too big now. Previously output cells seemed to truncate when you inserted the notebook into the chat context. For example, there was no guarantee that all visible output from df.head() in the notebook would end up in the chat. This new behavior probably makes more sense but some kind of toggle to exclude specific output datatypes might be nice. (or any mechanism to prevent me from blowing a bunch of tokens if I forget about an inline image)