Skip to content

fix: critical bugs, Gemini 3.5 Flash upgrade, and dependency updates - #1

Open
NanoBotAgent wants to merge 8 commits into
pyf-feifei:stablefrom
NanoBotAgent:fix/critical-bugs-and-deps
Open

fix: critical bugs, Gemini 3.5 Flash upgrade, and dependency updates#1
NanoBotAgent wants to merge 8 commits into
pyf-feifei:stablefrom
NanoBotAgent:fix/critical-bugs-and-deps

Conversation

@NanoBotAgent

Copy link
Copy Markdown

Summary

Critical bug fixes and dependency updates for ShortGPT:

Bug Fixes

  • Env var mismatch: config.py used ELEVEN_LABS_API_KEY but api_db.py and the .env template use ELEVENLABS_API_KEY — the ElevenLabs key could never be loaded from .env via the config module
  • Typo: Startup banner printed "STARTING SHORGPT" instead of "STARTING SHORTGPT" (2 instances)

Feature Updates

  • Gemini model upgrade: Replaced gemini-2.0-flash-lite-preview-02-05 with gemini-3.5-flash (stable, GA) for better quality and performance

Dependency Fixes

  • openai: Aligned requirements.txt (==1.37.0) with setup.py (==1.37.2), relaxed to >=1.37.2
  • httpx: Unpinned from ==0.27.2 to >=0.27.2 to avoid Gradio resolution conflicts
  • protobuf: Removed hardcoded 3.20.3 pin (outdated, not directly used)
  • pillow: Relaxed from ==10.4.0 to >=10.4.0 to allow security patches
  • gradio/gradio_client: Relaxed to >= minimum versions
  • Version bump: 0.1.31 → 0.1.32

Infrastructure

  • Dockerfile: Upgraded base image from python:3.10-slim-bullseye (Debian EOL) to python:3.10-slim-bookworm
  • .gitignore: Added **/__pycache__/ pattern

Files Changed

File Change
shortGPT/config/config.py Fix env var name ELEVEN_LABS_API_KEYELEVENLABS_API_KEY
shortGPT/gpt/gpt_utils.py Upgrade Gemini model to gemini-3.5-flash
gui/gui_gradio.py Fix "SHORGPT" → "SHORTGPT" typo
requirements.txt Relax pinned deps, remove protobuf pin
setup.py Align deps with requirements.txt, bump version
Dockerfile bullseye → bookworm
.gitignore Add **/__pycache__/

The config module used ELEVEN_LABS_API_KEY (with underscore) but
api_db.py and the .env template use ELEVENLABS_API_KEY (no underscore).
This mismatch meant the ElevenLabs key could never be loaded from .env
via the config module.
Replaced the outdated gemini-2.0-flash-lite-preview-02-05 with
gemini-2.5-flash for better quality and performance.
Replaced the outdated gemini-2.0-flash-lite-preview-02-05 with
gemini-3.5-flash for better quality, speed, and agentic capabilities.
Debian Bullseye reached end of life. Using bookworm for continued
security updates and package compatibility.
- openai: align with setup.py (>=1.37.2 instead of ==1.37.0)
- httpx: unpin to >=0.27.2 to avoid gradio resolution conflicts
- protobuf: remove hardcoded 3.20.3 pin (outdated, not directly used)
- pillow: relax to >=10.4.0 to allow security patches
- gradio/gradio_client: relax to >= minimum versions
- openai: >=1.37.2 (was ==1.37.2)
- pillow: >=10.4.0 (was ==10.4.0)
- Bump version to 0.1.32 for this bugfix release
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

Successfully merging this pull request may close these issues.

1 participant