Confusing (and destructive) behaviour on Tab or Return input
I most likely just activated some obscure setting by mistake, but I'm very confused with the behaviour I'm seeing at the moment and I'm unsure how to fix it. I recently (finally) upgraded to Rider 2025.1.4 from a 2021 perpetual licenese version and hadn't noticed this behaviour before the upgrade.
I have a simple base class that is currently inherited by one concrete class. When attempting to expand the base class with an additional method, hitting Enter (or Tab, for that matter) on any line in the code basically rewrites it, breaking it in the process.
This occurs with AI assistant disabled as well, so unless disabling isn't working as intended, this isn't assistant-related behaviour.
Two screenshots - the first shows the code and cursor position immediately before pressing return, while the second shows the resulting code and cursor position (removing both the interface inheritance and using statements):


This isn't the first time this has happened. I previously manually fixed it thinking I had made a mistake or that AI assistant was acting up, but has occurred enough times now to warrant looking into it a bit further.
Hoping someone else has had a similar issue (and more importantly, a simple permanent fix)…
Specs:
Rider 2025.1.4
Apple M2 Pro
macOS Sonoma 14.5
Please sign in to leave a comment.
Hello Chris,
Thanks for contacting Rider support. The symptom you observed does not look like Rider behavior. Could you try the following to revert Rider to initial state:
Clear Caches:
Reset IDE configuration
Follow Restore default settings to reset configurations
Disable All Downloaded Plugins:
If the issue continues
Thanks,
Tom
Thanks Tom - so far, resetting the IDE configuration seems to have done the trick (and has apparently eliminated some of the other strange issues I was having). If they reappear, I'll upload the logs as described.
I've seen similar issues in both PyCharm and Rider. Invalidating the cache appeared to solve it for some time, but the issue is coming back hunting me with any non-trivial projects. It is pretty annoying, because I end up repeatedly looking for “AI” settings I forgot to disable.
I think it is coming from the “full line auto-completion” feature (Settings: Editor / General / Inline completion) which I'm still using. It is a ~100MB model running entirely locally.
These “AI” features weren't introduced carefully enough, IMHO. They should not ever override the almost always correct, type system based auto-completion. They should not be enabled by default and definitely should not use any of the hotkeys people got used to invoke the type-system based auto-completion.
Likely related: https://youtrack.jetbrains.com/issue/LLM-18490/Cannot-prevent-AI-assitant-from-generating-code-on-Tab-press.
I agree 100%, and I do find myself fighting with the IDE in general (and more specifically, autocomplete) far more than I recall in my previous installation of Rider (2021). Autcomplete seems a bit better with the “Synchronize inline and popup completions” setting on (placing AI suggestions in the standard popup list), but in general, a lot of things just feel ‘off’ in a way I can't really put a finger on. I'm hoping these are just other settings adjustments I need to narrow in on…
However, I do believe my original issue was separate from these general annoyances and does appear to be a bug related to the settings I had from my previous installation. So far, resetting to defaults seems to have fixed that specific problem.