Moss Hope

Column embedding does exist. Adding clip_index I get ``` ERROR: pgvecto.rs: The given vector is invalid for input. ADVICE: Check if dimensions and scalar type of the vector is matched...

Altering the table and inserting the index got me going again too. From my logs (though I clipped a bit after I should have) and BlackHacks it looks like thats...

> This release added some missing validation to make config changes more robust. It had a side effect of exposing all the cases where something was wrong 😅 Makes sense....

> `clip_index` being completely missing isn't normal, so it wasn't accounted for. That's something that predates your change to a different model. But it would be nice if it created...

I'm not sure if this will be helpful for you or not. But by bisecting my vm backups I can say my clip_index was present on 5/27 running v1.101.0 and...

No problem. I'm not always the fastest at updating immich but the version numbers should at least clarify there On Mon, Aug 19, 2024, 11:22 AM Mert ***@***.***> wrote: >...

Arg! I have led you astray. I was looking back at release notes and saw the warning about v1.102.0 's databse changes. I did backup and restore to migrate my...

69610ec52429fecbe94c4c042cc42ab43e0491f8 is the first bad commit ``` commit 69610ec52429fecbe94c4c042cc42ab43e0491f8 Author: Autumn Ashton Date: Wed Jul 24 18:22:11 2024 +0100 steamcompmgr: Relative mouse fixes and improvements ```

It looks like `--force-grab-cursor` works. This still seems like a regression to me that should be addressed given it was working before. What do you think @misyltoad seems like `--force-grab-cursor`...

It could be, but it could be a different issue, the symptoms are somewhat different. Theirs might actually be from 3c5a232 , but probably not dade663