Skip to content

Add environment variable overrides for llama.cpp launches - #379

Merged
thomas9120 merged 1 commit into
mainfrom
beta
Sep 14, 2026
Merged

thomas9120 merged 1 commit into
mainfrom
beta

Conversation

@thomas9120

Copy link
Copy Markdown
Owner

Summary

Adds an advanced Environment Variables panel for configuring per-launch LLAMA_ and GGML_ overrides. Values are shared across Configure, Quick Launch, chat/model switching, benchmarking, memory estimation, and presets, then validated and applied only to the launched child process.

Scope

This does not support shell expansion or execution, LLAMA_GUI_/LLAMA_ARG_ settings, or environment-variable comparison and revert controls. Existing command-line argument controls remain the source of truth for GUI and CLI-managed options.

Verification

  • node tests/frontend/custom_launch_args_unit.cjs
  • Frontend flag-sync smoke coverage for field placement, validation, preset restoration, and launch request construction
  • tests/backend/test_process_environment.py covering validation, propagation, inherited values, runtime paths, and process-environment isolation
  • Updated module namespace, preset, and process lifecycle unit coverage

UI evidence

The Configure tab now includes an Environment Variables panel below Custom Launch Args, with inline validation and guidance for supported variable names and literal values.

- Validate and propagate LLAMA_/GGML_ settings across launch flows
- Add Configure UI, preset support, and coverage
@thomas9120
thomas9120 merged commit 0053c1b into main Sep 14, 2026
5 checks passed
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