[PR #1735] [MERGED] Stream subgraph output while it executes #2260

Closed
opened 2026-02-20 17:46:40 -05:00 by yindo · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/langchain-ai/langgraph/pull/1735
Author: @nfcampos
Created: 9/16/2024
Status: Merged
Merged: 9/17/2024
Merged by: @nfcampos

Base: mainHead: nc/16sep/stream-subgraph-in-progress


📝 Commits (5)

📊 Changes

6 files changed (+162 additions, -25 deletions)

View changed files

📝 libs/langgraph/langgraph/pregel/__init__.py (+10 -4)
📝 libs/langgraph/langgraph/pregel/runner.py (+37 -17)
libs/langgraph/langgraph/utils/aio.py (+35 -0)
📝 libs/langgraph/poetry.lock (+3 -3)
📝 libs/langgraph/pyproject.toml (+1 -1)
📝 libs/langgraph/tests/test_pregel_async.py (+76 -0)

📄 Description

  • previous behavior was to buffer all output from subgraph until it finished, now subgraph steps are emitted as soon as produced, while the subgraph is still running
  • this is slightly slower in benchmark scripts, but worth it as it's much "faster" in real-world latency

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/langchain-ai/langgraph/pull/1735 **Author:** [@nfcampos](https://github.com/nfcampos) **Created:** 9/16/2024 **Status:** ✅ Merged **Merged:** 9/17/2024 **Merged by:** [@nfcampos](https://github.com/nfcampos) **Base:** `main` ← **Head:** `nc/16sep/stream-subgraph-in-progress` --- ### 📝 Commits (5) - [`d163938`](https://github.com/langchain-ai/langgraph/commit/d163938a59261870b392bf164fe8d31cf41f6e59) Stream subgraph output while it executes - [`54207f4`](https://github.com/langchain-ai/langgraph/commit/54207f4d893bb507d950012cd53100efa811004f) Add test - [`c301198`](https://github.com/langchain-ai/langgraph/commit/c30119824c6261ab90c86682d27479ad70a55396) Add comment - [`bd2ecba`](https://github.com/langchain-ai/langgraph/commit/bd2ecba62240d68ee3262fc9949ccd963180e02c) Fix for py 3.9 - [`f59435a`](https://github.com/langchain-ai/langgraph/commit/f59435a892e96fb9092e0055a6df2e1dfc5111a9) Add more comments ### 📊 Changes **6 files changed** (+162 additions, -25 deletions) <details> <summary>View changed files</summary> 📝 `libs/langgraph/langgraph/pregel/__init__.py` (+10 -4) 📝 `libs/langgraph/langgraph/pregel/runner.py` (+37 -17) ➕ `libs/langgraph/langgraph/utils/aio.py` (+35 -0) 📝 `libs/langgraph/poetry.lock` (+3 -3) 📝 `libs/langgraph/pyproject.toml` (+1 -1) 📝 `libs/langgraph/tests/test_pregel_async.py` (+76 -0) </details> ### 📄 Description - previous behavior was to buffer all output from subgraph until it finished, now subgraph steps are emitted as soon as produced, while the subgraph is still running - this is slightly slower in benchmark scripts, but worth it as it's much "faster" in real-world latency --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
yindo added the pull-request label 2026-02-20 17:46:40 -05:00
yindo closed this issue 2026-02-20 17:46:40 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/langgraph#2260