Thanks Erick! This fixed that particular error, but now another popped up - i will mark your reply as ‘solution’, but allow me to share that new (but related) error here with you (i know technically I should start a new post, but i have been going in circles on this for days now).
NEW ERROR: “Error searching for documents: PGRST202 Could not find the function public.match_hkex_ipo_listings(filter, match_count, query_embedding)”
(there are other posts about this error in the forum, but i think my issue relates more to fundamental setup issue of my supabase table than a technical issue)
Error with Supabase Vector Stores)
NARROWING DOWN THE ERROR: I believe the error is caused by my embeddings column in the supabase table being empty: In Supabase i did the following steps:
- i created the table ‘hkex_ipo_listings’ and populated it manually via a CSV upload
- then i made the table vector enabled using the sql editor
- added the match fucntion for that table with the sql editor
- because i didnt upload the table content via n8n with an embedding model (but manually uploaded a CSV, i do have an embedding column in my table, BUT this embedding column is empty → do you think this is the issue: n8n’s supabase vector store has no vector embeddings to reference? (in my previous workflows were i uploaded a file via n8n the chunking and embedding happened ‘automaticlly’, so i was under the impression a supabase table would have that as a basic functionality’?)
Sorry again about sneaking in this ‘follow up quesiton’ and if the question isnt the smartest, Erick, but would appreicate your guidance on this a lot! I am really just trying to link the n8n vector node up to a supabase table (not a vectorised document, just a simple table like in excel -some numerical inputs, some text-) and I feel like it should be so easy.