Replies: 1 comment
|
Your approach is right in principle, and yes, both There's no build arg for extras in the Dockerfile ( Two things worth flagging on the specifics though. There is no langchain = [
"langsmith>=0.2.3,<1.0.0",
"langchain_text_splitters>=0.3.2,<1.0.0",
"langchain-core>=1.2.5"
]uv doesn't quietly ignore an undefined extra, it hard fails. So that build should already be erroring out unless something upstream is swallowing it, worth a look at your build logs.
aws = [
{ name = "s3fs", extra = ["boto3"] },
]
langchain = [
{ name = "langchain-core" },
{ name = "langchain-text-splitters" },
{ name = "langsmith" },
]so One caveat if what you're actually after is LangSmith tracing rather than just the If you only need the langchain packages importable, though, |
Uh oh!
There was an error while loading. Please reload this page.
I'm using Cognee in a Docker environment and need to add support for AWS and LangSmith integrations. I've extended the official Dockerfile by adding --extra aws --extra langsmith to both uv sync commands
https://github.com/topoteretes/cognee/blob/main/Dockerfile
Is this the recommended approach for extending the Dockerfile with additional extras, or should extras be configured differently for containerized deployments?
Any guidance would be appreciated!
All reactions