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.

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

Looking for Recommendations on how to Debug #23

Open
jjafuller opened this issue Nov 22, 2024 · 3 comments
Open

Looking for Recommendations on how to Debug #23

jjafuller opened this issue Nov 22, 2024 · 3 comments

Comments

@jjafuller
Copy link

I am wondering where I should get started trying to debug aider composer. I am attempting to use Ollama. From what I can gather it can send messages to the facade server, but there does not appear to be a call actually being made to the Ollama backend. I can see some log entries in the extension output, but they seem pretty high level.

2024-11-22 14:20:42.713 [info] Extension "aider-composer" is now active!
2024-11-22 14:20:42.713 [info] Starting aider-chat service...
2024-11-22 14:20:42.713 [info] aider-chat process args: C:\Projects\third-party\aider\.venv\Scripts\python.exe -m flask -A c:\Users\jjafc\.vscode\extensions\lee2py.aider-composer-1.3.0\server\main.py run --port 13491
2024-11-22 14:20:46.201 [info] aider-chat: WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
2024-11-22 14:20:46.201 [info] aider-chat:  * Running on http://127.0.0.1:13491
2024-11-22 14:20:46.202 [info] server started: http://127.0.0.1:13491
2024-11-22 14:20:46.202 [info] aider-chat: Press CTRL+C to quit
2024-11-22 14:20:48.842 [info] aider-chat: 127.0.0.1 - - [22/Nov/2024 14:20:48] "OPTIONS /api/chat/setting HTTP/1.1" 200 -
2024-11-22 14:20:48.845 [info] aider-chat: 127.0.0.1 - - [22/Nov/2024 14:20:48] "POST /api/chat/setting HTTP/1.1" 200 -
2024-11-22 14:20:52.294 [info] From Webview: sendChatMessage: hi
2024-11-22 14:20:52.313 [info] aider-chat: 127.0.0.1 - - [22/Nov/2024 14:20:52] "OPTIONS /api/chat HTTP/1.1" 200 -
2024-11-22 14:20:52.406 [info] aider-chat: 127.0.0.1 - - [22/Nov/2024 14:20:52] "POST /api/chat HTTP/1.1" 200 -
2024-11-22 14:21:24.185 [info] aider-chat: 127.0.0.1 - - [22/Nov/2024 14:21:24] "POST /api/chat HTTP/1.1" 200 -

This is the configuration I am using:

2024-11-22 15_08_47-● aider-composer md - n2y CapExEstimator - Visual Studio Code

I tried hitting the facade server locally to see if it returned any additional information.

curl --request POST \
  --url http://127.0.0.1:13491/api/chat \
  --header 'content-type: application/json' \
  --data '{
  "chat_type": "ask",
  "diff_format": "diff",
  "message": "hi",
  "reference_list": []
}'

This is the response, which doesn't give me much to go on.

event: usage
data: "Tokens: 68 sent, 0 received."

event: end

The following is the PowerShell script that I use to run aider in the terminal and it works as expected.

$env:OLLAMA_API_BASE = "http://192.168.5.15:11434" 
C:\\Projects\\third-party\\aider\\.venv\\Scripts\\aider.exe --model ollama/qwen2.5-coder:32b-instruct-q4_K_M

Is there a way to enable debug logging, or are there logs elsewhere I can look at?

@lee88688
Copy link
Owner

you can try latest version, it solve this problem.

@jmaerki
Copy link

jmaerki commented Nov 23, 2024

I had the same problem as jjafuller, and it started out the same after upgrading to 1.3.2, but then I just changed the LLM config again to a silly little "qwen2.5-coder:3b" and it started calling Ollama. However, it's now in an infinite loop producing the same unusable result each time and there's no button to cancel the request. ;-)

@lee88688
Copy link
Owner

I had the same problem as jjafuller, and it started out the same after upgrading to 1.3.2, but then I just changed the LLM config again to a silly little "qwen2.5-coder:3b" and it started calling Ollama. However, it's now in an infinite loop producing the same unusable result each time and there's no button to cancel the request. ;-)

In aider you can use ctrl+c to cancel, but in code, it seems there's no api to do this. I'll open an issue to aider to help.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants