langchain-ai/langgraph
1
0
Fork 0
mirror of https://github.com/langchain-ai/langgraph.git synced 2026-07-15 12:35:43 -04:00
Code Issues 314 Packages Projects Releases 17 Wiki Activity

[PR #5224] Draft : a filesystem checkpointer #4312

New Issue
Closed
opened 2026-02-20 17:50:00 -05:00 by yindo · 0 comments
yindo commented 2026-02-20 17:50:00 -05:00
Owner
Copy Link
Copy Source

Original Pull Request: https://github.com/langchain-ai/langgraph/pull/5224

State: closed
Merged: No

Hi folks,

I am experimenting with a new filesystem checkpointer, which I don't think exists in the codebase and didn't found in the community.

If there is already such a checkpointer, feel free to dismiss.

I am particularly interested into such a checkpointer because it can be coupled with Google Colab capability to mount Google Drive folders, so it would allow to save the agent state in a Google Drive folder.

This is draft pull-request.

Most of the changes at this point are AI-generated! (+1 hour of proof reading and tweaking things to understand what happens)

I am an experienced developer with some Python background but I don't practice this language, so I've used this PR as a practice and I had to figure new concepts I wasn't aware about (defaultdict for handling threads that do not exist yet, stacking exits, the PersistentDict implementation etc.).

  • At this point, it doesn't seem to actually create the file. I've pinpointed the issue to the fact that dumping files only happens in a PersistentDict close event. But I fail to understand when this happens, and whether I should dump the file more frequently eg on each "put" operation.
  • I think the stack exit calls should be reenabled in __enter, __exit and so on (they are present in the MemorySaver), but I am not 100% sure what they do so I'd be eager to get some confirmation (perhaps this explains why the file is not dumped)
  • The PR definitely lacks a unit test, I've just tested the code in a notebook by copy-pasting so far.
  • PersistentDict could be factored with InMemorySaver if we happen to reuse the exact same structure

So I hope this could help as a basis for a file system checkpointer, and I'd be eager to keep iterating on it with light directions from experienced contributors.

Also let me know if you prefer using a discussion rather than a pull request.

Update: after more exploration, I feel like PersistentDict is meant to be used as a InMemorySaver "factory" parameter, to allow dumping the data on save? So could InMemorySaver + PersistentDict act as a FileSystemSaver?

**Original Pull Request:** https://github.com/langchain-ai/langgraph/pull/5224 **State:** closed **Merged:** No --- Hi folks, I am experimenting with a new filesystem checkpointer, which I don't think exists in the codebase and didn't found in the community. If there is already such a checkpointer, feel free to dismiss. I am particularly interested into such a checkpointer because it can be coupled with Google Colab capability to mount Google Drive folders, so it would allow to save the agent state in a Google Drive folder. This is draft pull-request. **Most of the changes at this point are AI-generated!** (+1 hour of proof reading and tweaking things to understand what happens) I am an experienced developer with some Python background but I don't practice this language, so I've used this PR as a practice and I had to figure new concepts I wasn't aware about (`defaultdict` for handling threads that do not exist yet, stacking exits, the `PersistentDict` implementation etc.). - At this point, it doesn't seem to actually create the file. I've pinpointed the issue to the fact that dumping files only happens in a `PersistentDict` close event. But I fail to understand when this happens, and whether I should dump the file more frequently eg on each "put" operation. - I think the stack exit calls should be reenabled in __enter, __exit and so on (they are present in the MemorySaver), but I am not 100% sure what they do so I'd be eager to get some confirmation (perhaps this explains why the file is not dumped) - The PR definitely lacks a unit test, I've just tested the code in a notebook by copy-pasting so far. - PersistentDict could be factored with `InMemorySaver` if we happen to reuse the exact same structure So I hope this could help as a basis for a file system checkpointer, and I'd be eager to keep iterating on it with light directions from experienced contributors. Also let me know if you prefer using a discussion rather than a pull request. Update: after more exploration, I feel like `PersistentDict` is meant to be used as a `InMemorySaver` "factory" parameter, to allow dumping the data on save? So could `InMemorySaver` + `PersistentDict` act as a `FileSystemSaver`?
yindo added the pull-request label 2026-02-20 17:50:00 -05:00
yindo closed this issue 2026-02-20 17:50:00 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
agents
awaiting author revision
bug
change
chat models
checkpointer
ci
code quality
code quality
dependencies
documentation
drawing
duplicate
enhancement
good first issue
help wanted
invalid
maintainer
meta
missing information
not maintained by langchain
open-swe-auto
open-swe-max
open-swe-max-auto
pending
performance
pull-request
Mirrored from GitHub Pull Request
 question
requires design
sdk
server
stale
studio
tests
type checking
usage question
v0.5
v1.1
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/langgraph#4312
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?