I’m opening this request to say this would be a VERY HIGH VALUE feature for Cursor as most of us using this software are early adopters. I know Cursor is bleeding edge for AI powered development.
I believe an Easy Roll Back Button for us “Hard Heads” who love the “Most Recent” features and don’t want to be seen as only complainers would be a PERFECT SOLUTION to many problems.
Easy Roll Back + Transparent Communication (Add Agent to your status page Cursor! https://status.cursor.com/) Would make Cursor a TOP NOTCH product.
Customers who complain care.
Hear us out!
IF YOU CARE FOR CURSOR PLEASE VOTE FOR THIS FEATURE!!!
Thanks for sharing. I would comment more but don’t want to get this post flagged for advertising like someone else’s post got flagged and hidden earlier.
Hey Dan, I am speaking more specifically of Total Cursor App Easy Rollback Feature. Not within specific projects but for bleeding edge user like me who will always update to the most recent Cursor.
I am thinking one click functionality to move from 0.48.5 of Cursor back to 0.48.4 or 0.48.3
I would typically do this with my clients when a WordPress Version breaks their site and wait for a more stable version.
He is basically saying that every new release, the core feature of Cursor (aka your product market fit) is writing & applying code properly, breaks down. In fact for each new release Cursor seems to take 2 steps forward and 2.5 steps back. It’s frustrating!
For example, the launch of .46 was so unstable many people found out methods to reinstall .45. Then on .47 it stabilized for about two weeks with claude-3.7 sonnet Max, but now with the .48 rollout all the stability we rely upon seems to break. No real developer is really asking for “YOLO mode”.
Meanwhile all we see is that Cursor keeps closing fundraising rounds – as a founder I know that the Cursor founding team is focused on fundraising and not the core product.
All we ask is that cursor initiates an internal campaign to ensure that each new release has been thoroughly tested for UX before unleashing it upon your growing customer base. Then when we come back and complain, the burden is upon us to prove how bad it has become. This leads us to feel gaslit slightly.
Hey, we don’t really want to make a thing of supporting old versions, as doing so would slow down our development of new versions.
We aim to never remove a feature without replacing it with something better, and therefore aim to have everyone on the latest version and happy in doing so.
Is there something you’d look for in older versions that isn’t present in the latest?
Perhaps you guys can measure user satisfaction, retention, sentiment in different versions then see what went wrong, what went right with the new version?
Also I’ve never seen such a rapid version upgrade cycle for any product as I experience with Cursor. Maybe it’s okay to slow down the versioning to make sure it has a consistent behavior through time?
We feel jerked around with the different behaviors in each new upgrade (which we have to adjust our prompts and styles for each version). Especially true with devs with big projects.
thank you, I really dislike the new cursor. Specifically, its inability to parse a single large (200k) file. Cursor 0.45 can parse a single large file like a champ.
This isn’t about features, it’s about overall performance in that you guys often roll out newer updates that frankly ■■■■. I am so ■■■■■■ off that I upgraded from .47 to .48 or whatever this latest update is.
It ■■■■■ and gets stuck in coding loops now more than ever, seems to forget what i just told it to do and is back to ruining and changing old code that has nothing to do with the new featiure I am working on. Just takes it upon itself to go ahead and try to refactor everything it sees.
Many of us who use this product are not coders who can code from scratch. You guys MUST understand where your future is, not existing devs, as there are way more of us than existing developers, and frankly, I’d be running two different versions of your software if I were you.
Something like a “PRO” version focused entirely on existing Devs who need speed and other features that non-devs like me couldn’t care less about.
Then a “Basic” or “No-Code” version that focuses way more on accuracy and automated bug-fixing might run slower, but gets it right and doesn’t ravage your codebase and break your code. That’s all we care about. I need to build prototypes not production code.