Обсуждение: BUG #19114: ORDER BY ASC is tampering result when calculating distance btw vectors

The following bug has been logged on the website:

Bug reference:      19114
Logged by:          Naveen Krishna S
Email address:      naveenkrishna.s@sky.uk
PostgreSQL version: 14.13
Operating system:   Mac OS
Description:

In a table with vector column,
SELECT embedding <=> CAST('[0.01, 0.23, -0.1,..]' as vector)
AS distance
FROM my_table
WHERE TRUE
order by distance desc
LIMIT 100;

is giving 100 records whereas

SELECT embedding <=> CAST('[0.01, 0.23, -0.1,..]' as vector)
AS distance
FROM my_table
WHERE TRUE
order by distance asc
LIMIT 100;

is giving 40 records for a specific embedding and always less than the limit
for any query embedding. Why is this?
I have also noticed if I use NULLS FIRST or COALESCE(distance, 99999999999)
it is returning the requested limit. But when I tried to list the records
with distance as null, there were none.

On Mon, Nov 17, 2025 at 7:14 PM PG Bug reporting form
<noreply@postgresql.org> wrote:
> 40 records

Sounds like pgvector[1]?

"Why are there less results for a query after adding an HNSW index?
Results are limited by the size of the dynamic candidate list
(hnsw.ef_search), which is 40 by default. ...."

[1] https://github.com/pgvector/pgvector

1. The issue is occurring when HNSW indexing is done.
2. The issue is not replicable with vector extension 0.8.0
3. I have tried with multiple m and ef_construction pairs ranging from (16,200) to (64,512) and nothing is working.

Information in this email including any attachments may be privileged, confidential and is intended exclusively for the addressee. The views expressed may not be official policy, but the personal views of the originator. If you have received it in error, please notify the sender by return e-mail and delete it from your system. You should not reproduce, distribute, store, retransmit, use or disclose its contents to anyone. Please note we reserve the right to monitor all e-mail communication through our internal and external networks. SKY and the SKY marks are trademarks of Sky Limited and Sky International AG and are used under licence.

Sky UK Limited (Registration No. 2906991), Sky-In-Home Service Limited (Registration No. 2067075), Sky Subscribers Services Limited (Registration No. 2340150) and Sky CP Limited (Registration No. 9513259) are direct or indirect subsidiaries of Sky Limited (Registration No. 2247735). All of the companies mentioned in this paragraph are incorporated in England and Wales and share the same registered office at Grant Way, Isleworth, Middlesex TW7 5QD

On Wed, Nov 19, 2025 at 10:55 PM S, Naveen Krishna (Development
Engineer 3) <naveenkrishna.s@sky.uk> wrote:
> The issue is occurring when HNSW indexing is done.
> The issue is not replicable with vector extension 0.8.0
> I have tried with multiple m and ef_construction pairs ranging from (16,200) to (64,512) and nothing is working.
>
> May I know where could I see the release updates on 0.8.1 of the vector extension?
> Also please feel free to share your thoughts for me to test more on the same.

Hi Naveen,

pgvector is an independent project.  If you think it's misbehaving you
should probably raise an issue here (after reading all the
documentation):

https://github.com/pgvector/pgvector/issues