Pinecone-python-client v3.0.0 breaks pinecone-haystack

[awinml]

Describe the bug

The version 3.0.0 of pinecone-python-client is a big update and introduces a lot of changes. This new version is not compatible with the current implementation of pinecone-haystack and the code breaks as soon as you try to initialize the document store.

To Reproduce

The global pinecone.init has been removed, so pinecone-haystack gives an error as soon as you try to initialize the document store with the new version of pinecone-python-client.

Please see this Colab notebook for more information.

Immediate Fix

Pin pinecone-client to version 2.2.4 in the pinecone-haystack dependencies.

Describe the solution you'd like

Refactor pinecone-haystack to support the new version.

The new version introduces a lot of new features. Some of the changes include:

• Serverless indexes
• Global state variables set with pinecone.init have been refactored into new Pinecone class instances that encapsulate their configuration state.
• Everything GRPC-related is now moved into a subpackage, pinecone.grpc, so that GRPC code is only imported when needed.
• The list_indexes and list_collections methods now return an array with full descriptions of each resource, not merely an array of names.
• The Pinecone SDK now has a Apache 2.0 license.

Please see the Migration Guide for version 3.0.0 for more information.

Describe your environment:

• OS: Linux
• Haystack version: haystack-ai - 2.0.0-beta.4
• Integration version: pinecone-haystack - 0.0.1

[anakin87]

Thank you for reporting this!

I would adopt the immediate fix and then refactor the Document Store.

[masci]

0.1.0 was released to ship the temporary workaround

[bilgeyucel]

We can resolve #715 with this upgrade

[mmurraysans]

Hi All,

I took a pass at updating this to the new client in #756

[anakin87]

TODO:

• Upgrade the Document Store and Retriever (feat!: Pinecone - support for the new API #793)
• Update examples (feat!: Pinecone - support for the new API #793)
• The package has been released on PyPI
• Docs at https://docs.haystack.deepset.ai/ were updated (in review...)
• Integration tile on https://github.com/deepset-ai/haystack-integrations was updated (update Pinecone integration page based on the new API haystack-integrations#240)