Adopt command_handler crate, upgrade llama-cpp-bindings to 0.9.0, and remove inline imports/lint suppressions

[mcharytoniuk]

Adopts the published command_handler crate (Handler, shutdown signal, socket-addr value parser), upgrades to llama-cpp-bindings 0.9.0, and removes inline imports, pub use re-exports, and in-code clippy suppressions.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR modernizes the workspace by adopting the external command_handler crate for common CLI/app concerns, upgrading llama-cpp-bindings to 0.9.0, and removing a large amount of inline-import and clippy-suppression noise. It also updates GPU offload layer counts to support “all layers” semantics via -1, and improves robustness around token classification/detokenization failures in the continuous-batch scheduler.

Changes:

• Adopt command_handler for the Handler trait, shutdown-signal handling, and socket-addr parsing; remove local equivalents.
• Upgrade llama-cpp-bindings* to =0.9.0 and adjust call sites (e.g., validator_for, MTMD marker, sampler error handling).
• Refactor/import cleanup across tests and crates, plus introduce a new terminal error token (DetokenizationFailed) that is treated as “done”.

Reviewed changes

Copilot reviewed 111 out of 112 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
paddler_tests/tests/qwen35_internal_endpoint_with_thinking_enabled_emits_reasoning_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen35_internal_endpoint_with_thinking_disabled_emits_only_content_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen35_internal_endpoint_emits_tool_call_parsed_event.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/qwen35_internal_endpoint_emits_reasoning_tokens_for_image_request.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_responses_non_streaming_returns_text_and_usage.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_openai_streaming_usage_breakdown_with_thinking.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_openai_streaming_emits_usage_when_requested.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_openai_streaming_emits_tool_calls_for_function_tool.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_openai_non_streaming_usage_with_tool_calls.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_openai_non_streaming_returns_usage.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_openai_non_streaming_emits_tool_calls_for_function_tool.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_internal_endpoint_with_thinking_enabled_emits_reasoning_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_internal_endpoint_with_thinking_disabled_emits_no_reasoning_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_internal_endpoint_tools_without_parse_flag_emit_only_raw_tokens.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/qwen3_internal_endpoint_pure_content_usage.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_internal_endpoint_max_tokens_usage_matches.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/qwen3_internal_endpoint_emits_tool_call_tokens.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/qwen3_internal_endpoint_emits_tool_call_parsed_event.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/qwen3_internal_endpoint_concurrent_requests_independent_usage.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/mistral3_internal_endpoint_emits_tool_call_parsed_event.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/mistral3_internal_endpoint_emits_reasoning_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/mistral3_internal_endpoint_emits_reasoning_tokens_for_image_request.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/gemma4_internal_endpoint_emits_tool_call_parsed_event.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/gemma4_internal_endpoint_emits_reasoning_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/gemma4_internal_endpoint_emits_reasoning_tokens_for_image_request.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/deepseek_r1_distill_llama_8b_internal_endpoint_emits_reasoning_tokens.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/continuous_batch_stops_generation_when_stop_sender_dropped.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/continuous_batch_releases_slots_on_shutdown_with_active_request.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/continuous_batch_rejects_embedding_during_active_generation.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/continuous_batch_generates_tokens_with_partial_layer_offload.rs	Align partial GPU layer count type with i32.
paddler_tests/tests/agent_reports_tool_call_validation_failure.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/agent_reports_grammar_initialization_failure_for_invalid_gbnf.rs	Import cleanup and simplified anyhow error construction.
paddler_tests/tests/agent_rejects_tool_with_invalid_required_field_in_schema.rs	Use json! macro import and simplify.
paddler_tests/tests/agent_rejects_structurally_invalid_tool_schema.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/agent_pipeline_recognizes_duck_typed_tool_call_format_when_template_is_not_registered.rs	Import cleanup; use json! and simplified anyhow errors.
paddler_tests/tests/agent_conversation_with_function_tool_succeeds.rs	Use json! macro import.
paddler_tests/src/model_card/mod.rs	Change GPU layer count type to i32.
paddler_test_cluster_harness/src/wait_until_healthy.rs	Simplify anyhow error creation.
paddler_test_cluster_harness/src/resource_snapshot.rs	Narrow std::fs imports to used items.
paddler_test_cluster_harness/src/load_test_image_data_uri.rs	Narrow std::fs imports to used items.
paddler_test_cluster_harness/src/collect_generated_tokens.rs	Test cleanup and simplified anyhow error creation.
paddler_test_cluster_harness/src/collect_embedding_results.rs	Test cleanup and simplified anyhow error creation.
paddler_test_cluster_harness/src/buffered_requests_stream_watcher.rs	Test cleanup and simplified anyhow error creation.
paddler_openai_response_format_validator/src/openai_validator.rs	Use validator_for import and simplify call site.
paddler_messaging/src/inference_parameters.rs	n_gpu_layers type change to i32 and document -1 semantics.
paddler_messaging/src/generated_token_result.rs	Add DetokenizationFailed terminal outcome and tests.
paddler_messaging/src/embedding_normalization_method.rs	Narrow std::mem import to discriminant.
paddler_gui/src/app.rs	Switch shutdown-signal registration to command_handler.
paddler_gui/Cargo.toml	Add command_handler dependency.
paddler_download_manager/tests/download.rs	Narrow tokio::fs imports and update call sites.
paddler_download_manager/src/stream_to_partial_file.rs	Narrow tokio imports in tests and update call sites.
paddler_download_manager/src/partial_file.rs	Narrow tokio::fs imports and update call sites/tests.
paddler_client/src/inference_socket/spawn_read_task.rs	Use debug! macro via direct import.
paddler_client/src/error.rs	Remove inline clippy suppression (policy moved to workspace lints).
paddler_cli/src/main.rs	Make main explicit and call run() via imports.
paddler_cli/src/lib.rs	Adopt command_handler handler/shutdown integration; box Balancer subcommand.
paddler_cli/src/cmd/value_parser/parse_socket_addr.rs	Delegate socket resolution to command_handler.
paddler_cli/src/cmd/mod.rs	Remove local handler module export.
paddler_cli/src/cmd/handler.rs	Remove local Handler trait (replaced by command_handler).
paddler_cli/src/cmd/balancer.rs	Implement external Handler trait; adjust async-trait settings/signature.
paddler_cli/src/cmd/agent.rs	Implement external Handler trait; adjust async-trait settings/signature.
paddler_cli/Cargo.toml	Add command_handler; remove now-unused deps.
paddler_cli_tests/src/model_card/mod.rs	Change GPU layer count type to i32.
paddler_cache_dir/src/lib.rs	Select platform-specific cache_dir module via #[path].
paddler_cache_dir/src/cached_downloaded_model.rs	Narrow tokio::fs imports to used items.
paddler_cache_dir/src/cache_dir/windows.rs	Narrow std::env import and simplify test import.
paddler_cache_dir/src/cache_dir/unix.rs	Narrow std::env import and simplify test import.
paddler_cache_dir/src/cache_dir/mod.rs	Remove old cfg re-export shim (replaced by crate-root module selection).
paddler_bootstrap/src/shutdown_signal/windows.rs	Remove internal shutdown-signal implementation (moved to command_handler).
paddler_bootstrap/src/shutdown_signal/unix.rs	Remove internal shutdown-signal implementation (moved to command_handler).
paddler_bootstrap/src/shutdown_signal/mod.rs	Remove internal shutdown-signal module export.
paddler_bootstrap/src/service_thread.rs	Narrow std::thread import and adjust JoinHandle type usage.
paddler_bootstrap/src/lib.rs	Stop exporting removed shutdown_signal module.
paddler_bootstrap/Cargo.toml	Remove unix-only dev-dependency no longer needed.
paddler_balancer/src/web_admin_panel_service/http_route/static_files.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/web_admin_panel_service/http_route/home.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/web_admin_panel_service/http_route/favicon.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/statsd_service/mod.rs	Remove inline lint suppression (policy moved to workspace lints).
paddler_balancer/src/state_database/file/mod.rs	Narrow tokio::fs imports and update call sites/tests.
paddler_balancer/src/response/view_from_http_response_builder.rs	Narrow std::mem usage to discriminant and adjust assertions.
paddler_balancer/src/management_service/mod.rs	Make test helpers fallible rather than unwrap-based.
paddler_balancer/src/management_service/http_route/api/ws_agent_socket/mod.rs	Remove inline lint suppression (policy moved to workspace lints).
paddler_balancer/src/management_service/http_route/api/put_balancer_desired_state.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/management_service/http_route/api/get_buffered_requests.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/management_service/http_route/api/get_balancer_desired_state.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/management_service/http_route/api/get_balancer_applicable_state.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/management_service/http_route/api/get_agents.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/inference_service/http_route/api/ws_inference_socket/mod.rs	Remove inline lint suppressions; narrow actix-web test imports.
paddler_balancer/src/inference_service/http_route/api/post_generate_embedding_batch.rs	Narrow actix-web test imports to used items.
paddler_balancer/src/controls_websocket_endpoint.rs	Remove inline lint suppression (policy moved to workspace lints).
paddler_balancer/src/compatibility/openai_service/responses_streaming_response_transformer.rs	Remove inline lint suppression (policy moved to workspace lints).
paddler_balancer/src/compatibility/openai_service/responses_response_builder.rs	Use json! macro import in tests.
paddler_balancer/src/compatibility/openai_service/openai_message.rs	Use json! macro import in tests.
paddler_balancer/src/compatibility/openai_service/openai_error.rs	Treat DetokenizationFailed as an error token for OpenAI mapping; test import cleanup.
paddler_balancer/src/compatibility/openai_service/openai_completion_request_params.rs	Use json! macro import in tests.
paddler_balancer/src/compatibility/openai_service/http_route/post_chat_completions.rs	Use json! macro import in tests.
paddler_balancer/src/compatibility/openai_service/arguments_to_tool_call_string.rs	Use json! macro import in tests.
paddler_balancer/src/agent_controller_pool.rs	Simplify anyhow error construction.
paddler_agent/src/tool_call_validator.rs	Use validator_for import; json! macro import in tests.
paddler_agent/src/tool_call_buffer.rs	Narrow std::mem import to take.
paddler_agent/src/sample_token_at_batch_index.rs	Add error context when applying samplers.
paddler_agent/src/prepare_conversation_history_request.rs	Update MTMD marker creation to fallible API (mtmd_default_marker()?).
paddler_agent/src/model_source/url.rs	Narrow tokio::fs imports in tests and update call sites.
paddler_agent/src/model_source/local.rs	Narrow tokio::fs import to try_exists.
paddler_agent/src/decoded_image.rs	Import narrowing and small refactors; use direct imports for image/resvg helpers.
paddler_agent/src/continuous_batch_scheduler/mod.rs	Make token classifier construction fallible; handle detokenization failures as terminal outcomes.
paddler_agent/src/continuous_batch_scheduler/classify_token_phase.rs	Propagate classifier ingest errors via Result.
paddler_agent/src/continuous_batch_scheduler/advance_generating_phase.rs	Convert classifier errors into DetokenizationFailed completion.
paddler_agent/src/chat_template_renderer/pyjinja_tojson.rs	Remove inline lint suppression (policy moved to workspace lints).
paddler_agent/src/chat_template_renderer/mod.rs	Import add_to_environment directly.
Cargo.toml	Add command_handler; upgrade llama-cpp-bindings*; adjust workspace clippy lint policy.
Cargo.lock	Lock command_handler and upgraded llama-cpp-bindings* dependency graph.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.