Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

L: Extensions must configure autocomplete provider client-side even if proxied through Sourcegraph instance #931

Closed
Tracked by #56161
chwarwick opened this issue Sep 5, 2023 · 2 comments
Closed
Tracked by #56161

L: Extensions must configure autocomplete provider client-side even if proxied through Sourcegraph instance #931

chwarwick opened this issue Sep 5, 2023 · 2 comments
Assignees
@taras-yemets

Comments

@chwarwick
Copy link
Contributor

chwarwick commented Sep 5, 2023 •
edited
Loading

Currently the autocomplete provider in the extension must be manually selected if a user is not using the default Anthropic provider.

Sourcegraph instances can be be configured to use the following providers for autocomplete:

  • Sourcegraph gateway
  • Anthropic (using own keys)
  • Azure OpenAI (using own keys)

When an extensions connects to a sourcegraph instance the correct autocomplete provider should be defaulted based on the how the instance is configured. For example if the sourcegraph instance is using the Azure OpenAI provider the extension should default to using the unstable-openai
@chwarwick chwarwick mentioned this issue Sep 5, 2023
Closed
@chwarwick
Copy link
Contributor Author

Some potential solutions:

  • Extending the current feature flag based solution to work for instances that are not sourcegraph.com

I think this is a reasonable approach if we end up in a situation where we need a quick solution but isn't ideal

  • Extending the current logic that gets the LLMConfiguration to include enough information to properly select an autocomplete provider. To me this feels like a better long term solution.

To properly select the extension autocomplete provider the site completion provider and model are potentially needed. When using the sourcegraph provider the model will be prefixed with the type openai, anthropic etc... If using Azure OpenAI the provider is the relevant piece of information as the model name is user defined and therefore isn't useful.

This was referenced Sep 13, 2023
Merged
Merged
taras-yemets added a commit that referenced this issue Sep 15, 2023
@taras-yemets
configure autocomplete provider based on cody LLM settings in site co…
c475300 
…nfig (#1035)

Part of #931
Test in pair with sourcegraph/sourcegraph#56568

Defines the autocomplete provider config based on the completions
provider and model names from the site config.
The suggested configuration hierarchy:
1. If `cody.autocomplete.advanced.provider` field in VSCode settings is
set to a supported provider name, and all the additional conditions like
model, access token, etc. are met the corresponding provider config is
returned. Otherwise, return `null` (completions provider is not
created).
2. If the provider name and model can be defined based on the evaluated
feature flags, return the corresponding provider.
3. If the completions provider name is defined in the connected
Sourcegraph instance site config, we return the corresponding provider
config. If the provider name/model can't be parsed or this provider is
not supported `null` is returned (completions provider is not created).
4. Anthropic config provider is used by default.
@taras-yemets
Copy link
Contributor

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@taras-yemets taras-yemets

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@chwarwick @taras-yemets