Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Improve vector store API#1201

Merged
malatewang merged 24 commits into
MemMachine:mainfrom
edwinyyyu:batch_vectors
Apr 2, 2026
Merged

Improve vector store API#1201
malatewang merged 24 commits into
MemMachine:mainfrom
edwinyyyu:batch_vectors

Conversation

@edwinyyyu
Copy link
Copy Markdown
Contributor

@edwinyyyu edwinyyyu commented Mar 9, 2026 •
edited
Loading

Motivation:

  • Update APIs to be easier to read, work with, and extend.
  • Support more efficient operations.
  • Define concurrency models for horizontal scaling.

Changes:

  • query now returns an Iterable with one item per query vector.
  • Add collection startup and shutdown for cases where collection must be loaded.
  • Support multitenancy via sharding and define concurrency guarantees.
  • Fix default booleans to match docstring comments.
  • Naming constraints.

Details:

  • Most vector databases recommend against one collection per tenant, and empirical tests show that many collections induces significant overhead.
  • We guarantee safe concurrent usage on a single process.
  • We guarantee safe concurrent sharded usage (one node per logical collection) on multiple processes.

Naming constraints determined by lowest common denominators for support.

@edwinyyyu edwinyyyu changed the title Improve vector query return value Improve vector store API based on discussions Mar 10, 2026
@edwinyyyu edwinyyyu marked this pull request as draft March 11, 2026 18:43
@edwinyyyu edwinyyyu mentioned this pull request Mar 11, 2026
Closed
26 tasks
@edwinyyyu edwinyyyu changed the title Improve vector store API based on discussions Improve vector store API Mar 13, 2026
@edwinyyyu
Copy link
Copy Markdown
Contributor Author

edwinyyyu commented Mar 13, 2026

Blocks #1199, #1205

@edwinyyyu edwinyyyu marked this pull request as ready for review March 13, 2026 19:54
@edwinyyyu edwinyyyu mentioned this pull request Mar 14, 2026
Merged
26 tasks
@edwinyyyu edwinyyyu force-pushed the batch_vectors branch 3 times, most recently from b3b2a57 to cf599b4 Compare March 17, 2026 23:37
malatewang
malatewang requested changes Mar 17, 2026
View reviewed changes
Comment thread packages/server/src/memmachine_server/common/vector_store/vector_store.py Outdated
async def upsert(
self,
*,
partition_key: str,
Copy link
Copy Markdown
Contributor

@malatewang malatewang Mar 14, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why do we need partition here?

Copy link
Copy Markdown
Contributor Author

@edwinyyyu edwinyyyu Mar 18, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fixed.

Comment thread packages/server/src/memmachine_server/common/vector_store/vector_store.py
) -> None:
"""
Create a collection in the vector store.
Create a logical collection in the vector store and return a handle to it.
Copy link
Copy Markdown
Contributor

@malatewang malatewang Mar 17, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The return type should be Collection

Copy link
Copy Markdown
Contributor Author

@edwinyyyu edwinyyyu Mar 18, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In open_or_create_collection.

Comment thread packages/server/src/memmachine_server/common/vector_store/vector_store.py Outdated

@abstractmethod
async def delete_collection(self, collection_name: str) -> None:
async def open_collection(self, *, namespace: str, name: str) -> Collection:
Copy link
Copy Markdown
Contributor

@malatewang malatewang Mar 17, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

return type should be Collection | None

Copy link
Copy Markdown
Contributor Author

@edwinyyyu edwinyyyu Mar 18, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fixed.

@edwinyyyu edwinyyyu requested a review from malatewang March 18, 2026 00:53
@edwinyyyu
Copy link
Copy Markdown
Contributor Author

edwinyyyu commented Mar 18, 2026

Before merge, need to check that (de)serialization is deterministic.

@edwinyyyu
Copy link
Copy Markdown
Contributor Author

edwinyyyu commented Mar 18, 2026 •
edited
Loading

Before merge, need to check that (de)serialization is deterministic.

Tests work now on Qdrant implementation branch.

@edwinyyyu edwinyyyu force-pushed the batch_vectors branch 3 times, most recently from d69686f to 69fa649 Compare March 27, 2026 19:24
@edwinyyyu edwinyyyu mentioned this pull request Apr 1, 2026
Merged
26 tasks
edwinyyyu added 18 commits April 1, 2026 22:14
@edwinyyyu
Reapply "Hoist partition key concept up -- now namespaces and names"
19aec8e 
This reverts commit ba3dcb3.
@edwinyyyu
Define concurrency guarantees
b51a62a 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Revert two commits
12a9146 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Revert "Revert two commits"
a12e15c 
This reverts commit 08ecc96.
@edwinyyyu
Update concurrency guarantees
a6257f6 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Update lifecycle APIs
e3d77e5 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Remove unnecessary error
aef722a 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Add new error
a4b1615 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Add collection configuration data model
491afd4 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Rename error for symmetry and simplicity
74da1c4 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Guarantee serialization order
3be75a1 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Support deserialization
0b27fef 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Fix return type
031aa6e 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Don't confuse {} and None
bb18f3b 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Add naming constraints
d6ff783 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Validate identifiers
1dbb8ab 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Add timezone utilities
dc52d85 
Signed-off-by: Edwin Yu <[email protected]>
@edwinyyyu
Define behavior for empty Record attributes
63dfe14 
Signed-off-by: Edwin Yu <[email protected]>
@malatewang malatewang merged commit a7c2d0e into MemMachine:main Apr 2, 2026
44 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@malatewang malatewang malatewang approved these changes

@jealous jealous Awaiting requested review from jealous

@o-love o-love Awaiting requested review from o-love

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@edwinyyyu @malatewang