emozilla
43e67d872f
Run models locally as a first-class provider. The CLI grows a managed llama.cpp runtime (engine install, model download, server supervision); the desktop app grows the full setup and management story on top of it. GUI surfaces ship behind the desktop --local launch flag (hermes desktop --local, or the flag on the packaged app); backend routes and the CLI are always live. Runtime (hermes_cli/local_runtime/): - curated GGUF catalog with per-machine variant selection: hardware probe (VRAM/RAM/UMA), fit planning with spill accounting, quant choice by context window - derived recommendation: quality-ranked picks gated by a predicted decode-speed floor, bandwidth-aware on unified memory; the decision table is pinned as a test (pick AND reason per memory class), and the Recommended badge explains its pick in a tooltip fed by the resolver's actual branch - engine install + model download with resumable split parts, cumulative plan-level progress, and staged-model integrity (a split GGUF counts only when every part is present) - server supervision: spawn/adopt/stop, router mode with per-model load progress relayed over SSE, abandoned-request cleanup Desktop: - Settings -> Providers -> Local models: one-click quickstart (install engine, download the recommended model, boot) plus per-model download/ activate/eject, fit-ranked catalog with context pills - model pickers (composer dropdown + Cmd+K) show staged local models, in-flight downloads as live progress rows, and load-into-memory bars - local-setup campaign tip for eligible hardware; System resources statusbar widget (GPU/VRAM/RAM); in-chat load progress during sends - friendly dead-server errors, and failed agent builds retry on the next send instead of wedging the session Co-developed with NVIDIA field feedback on RTX 5090 and DGX Spark.
124 lines
4.5 KiB
Python
124 lines
4.5 KiB
Python
"""Engine-update contracts (Rollout 4 follow-up):
|
|
|
|
- default tag flows from DEFAULT_CONFIG unless the user pinned;
|
|
- boot serves what is INSTALLED, never downloads (the ladder);
|
|
- update_available only when the local engine is enabled AND installed
|
|
AND the configured tag is missing on disk;
|
|
- the update itself is a button-driven job, and prune keeps N-1.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import json
|
|
|
|
import pytest
|
|
|
|
|
|
@pytest.fixture
|
|
def hermes_home(tmp_path, monkeypatch):
|
|
home = tmp_path / ".hermes"
|
|
home.mkdir()
|
|
monkeypatch.setenv("HERMES_HOME", str(home))
|
|
return home
|
|
|
|
|
|
def _install_fake_tag(home, tag: str, backend: str = "cuda") -> None:
|
|
d = home / "runtimes" / "llamacpp" / tag / backend
|
|
d.mkdir(parents=True)
|
|
(d / "manifest.json").write_text(json.dumps({
|
|
"tag": tag, "backend": backend, "assets": {},
|
|
"verified_version": f"version: {tag.lstrip('b')}",
|
|
}), encoding="utf-8")
|
|
# server_binary() looks for the executable name per-OS; give it both.
|
|
(d / "llama-server.exe").write_bytes(b"MZ fake")
|
|
(d / "llama-server").write_bytes(b"\x7fELF fake")
|
|
|
|
|
|
def test_installed_tags_newest_first(hermes_home):
|
|
from hermes_cli.local_runtime.binaries import installed_tags
|
|
|
|
assert installed_tags() == []
|
|
_install_fake_tag(hermes_home, "b10290")
|
|
_install_fake_tag(hermes_home, "b10412")
|
|
assert installed_tags() == ["b10412", "b10290"]
|
|
|
|
|
|
def test_default_tag_flows_from_default_config(hermes_home):
|
|
"""Unpinned users inherit the Hermes-release default (deep-merge);
|
|
the shipped default must be a plausible rolling tag."""
|
|
from hermes_cli.config import load_config
|
|
from hermes_cli.config_defaults import DEFAULT_CONFIG
|
|
|
|
default_tag = DEFAULT_CONFIG["local_runtime"]["tag"]
|
|
assert default_tag.startswith("b") and default_tag.lstrip("b").isdigit()
|
|
assert load_config()["local_runtime"]["tag"] == default_tag
|
|
|
|
|
|
def test_update_available_requires_enabled_and_installed(hermes_home, monkeypatch):
|
|
"""The flag's truth table: enabled+installed+configured-missing only."""
|
|
from fastapi.testclient import TestClient
|
|
|
|
from hermes_cli import web_server
|
|
|
|
client = TestClient(web_server.app)
|
|
# Same auth pattern as the other local-models route tests.
|
|
client.headers[web_server._SESSION_HEADER_NAME] = web_server._SESSION_TOKEN
|
|
|
|
def status():
|
|
r = client.get("/api/local-models/status")
|
|
assert r.status_code == 200, r.text
|
|
return r.json()
|
|
|
|
import hermes_cli.web_routers.local_models as lm
|
|
|
|
# Case 1: enabled, configured newer than installed -> update available.
|
|
monkeypatch.setattr(lm, "_runtime_section",
|
|
lambda: {"enabled": True, "tag": "b10412"})
|
|
_install_fake_tag(hermes_home, "b10290")
|
|
s = status()
|
|
assert s["update_available"] is True
|
|
assert s["configured_tag"] == "b10412"
|
|
assert s["tag"] == "b10290" # serving what's installed
|
|
|
|
# Case 2: configured tag installed -> no update.
|
|
_install_fake_tag(hermes_home, "b10412")
|
|
s = status()
|
|
assert s["update_available"] is False
|
|
assert s["tag"] == "b10412"
|
|
|
|
# Case 3: disabled -> never flagged, even with a mismatch.
|
|
monkeypatch.setattr(lm, "_runtime_section",
|
|
lambda: {"enabled": False, "tag": "b10999"})
|
|
assert status()["update_available"] is False
|
|
|
|
|
|
def test_boot_never_downloads_missing_tag(hermes_home, monkeypatch):
|
|
"""The ladder: configured-but-not-installed serves the newest installed
|
|
tag; nothing installed means no boot (and NO download either way)."""
|
|
from hermes_cli.local_runtime import bootstrap
|
|
|
|
calls = []
|
|
monkeypatch.setattr(
|
|
"hermes_cli.local_runtime.binaries.ensure_runtime_installed",
|
|
lambda tag, backend, **kw: calls.append(tag) or (_ for _ in ()).throw(
|
|
AssertionError("boot must not reach install for missing tags")))
|
|
|
|
# Nothing installed: returns None before any install attempt.
|
|
cfg = {"local_runtime": {"enabled": True, "tag": "b10412"}}
|
|
assert bootstrap.ensure_local_runtime(cfg) is None
|
|
assert calls == []
|
|
|
|
|
|
def test_prune_keeps_n_minus_one(hermes_home):
|
|
from hermes_cli.local_runtime.binaries import installed_tags, prune_old_tags
|
|
|
|
for tag in ("b10100", "b10200", "b10290"):
|
|
_install_fake_tag(hermes_home, tag)
|
|
prune_old_tags(["b10290", "b10200"])
|
|
assert installed_tags() == ["b10290", "b10200"]
|
|
# downloads/ cache dir must survive pruning when present.
|
|
downloads = hermes_home / "runtimes" / "llamacpp" / "downloads"
|
|
downloads.mkdir(exist_ok=True)
|
|
prune_old_tags(["b10290"])
|
|
assert downloads.exists()
|