mirror of
https://github.com/lancedb/lancedb.git
synced 2026-10-15 00:02:26 +00:00
Jack Ye
b21fb2a4dc
The pages describing this code lived in lancedb/docs, a repository with ten stars that neither contributing guide mentioned. A user who hit a wrong page while reading the source had to find a separate, near-invisible repository to fix it. These 102 pages now sit beside the code, and a documentation fix is a pull request in the repository the reader already had open. `docs/web` is a complete Mintlify site: `cd docs/web && mint dev` serves it with no other checkout. Enterprise pages stay private and are merged in when the published site is assembled, so links into them do not resolve here — every open-source page does. `docs/web-tests` holds the tests the code examples are extracted from. They ran against a published wheel before; here they can run against this repository's own SDK, so an example that stops compiling fails the pull request that broke it rather than being noticed later. Both directories sit beside `docs/src` and are invisible to it: mkdocs reads `docs_dir: src`, so the SDK reference builds and deploys exactly as before.
23 lines
4.1 KiB
Plaintext
23 lines
4.1 KiB
Plaintext
{/* Auto-generated by scripts/mdx_snippets_gen.py. Do not edit manually. */}
|
|
|
|
export const PyBoundedCache = "from functools import lru_cache\n\nclass GoodEmbedding:\n def __init__(self):\n self._embed = None\n\n def setup(self):\n model = load_model()\n self._embed = lru_cache(maxsize=1024)(model.embed)\n\n def __call__(self, text: str) -> list[float]:\n if self._embed is None:\n self.setup()\n return self._embed(text)\n";
|
|
|
|
export const PyConfidenceCheck = "def __call__(self, x):\n scratch = bytearray(8 * 1024 * 1024) # 8 MiB\n self._scratches.append(scratch) # <-- deliberate leak\n return ...\n";
|
|
|
|
export const PyLeakyAggregator = "class BadAggregator:\n def __init__(self):\n self.history = []\n\n def __call__(self, batch: pa.RecordBatch) -> pa.Array:\n self.history.append(batch) # holds every batch ever processed\n ...\n";
|
|
|
|
export const PyLeakyCache = "class BadEmbedding:\n def __init__(self):\n self.cache: dict[str, list[float]] = {}\n\n def __call__(self, text: str) -> list[float]:\n if text not in self.cache:\n self.cache[text] = self.model.embed(text)\n return self.cache[text]\n";
|
|
|
|
export const PyLeakyClosure = "class BadDeferred:\n def __init__(self):\n self.work_queue = []\n\n def __call__(self, x: pa.Array) -> pa.Array:\n # Lambda captures `x` by reference — the whole Array stays alive\n self.work_queue.append(lambda: expensive(x))\n ...\n";
|
|
|
|
export const PyLogMemory = "import resource, pyarrow as pa\n\ndef log_memory(seq: int) -> None:\n rss_bytes = resource.getrusage(resource.RUSAGE_SELF).ru_maxrss\n # ru_maxrss is bytes on macOS, KiB on Linux:\n import sys\n if sys.platform != \"darwin\":\n rss_bytes *= 1024\n arrow_live = pa.total_allocated_bytes()\n print(\n f\"seq={seq} \"\n f\"rss_mb={rss_bytes // 1024**2} \"\n f\"arrow_live_mb={arrow_live // 1024**2} \"\n f\"gap_mb={(rss_bytes - arrow_live) // 1024**2}\",\n flush=True,\n )\n";
|
|
|
|
export const PyMemrayTrackerUdf = "import os, pathlib, uuid\nfrom typing import Any\nimport memray\nimport geneva\nimport pyarrow as pa\n\n_MEMRAY_OUT_DIR_ENV = \"MY_UDF_MEMRAY_OUT_DIR\"\n\n\n@geneva.udf(data_type=pa.list_(pa.float32(), 512))\nclass MyEmbedding:\n def __init__(self):\n self.model = None\n self._tracker: Any = None # memray.Tracker, when profiling is on\n\n def setup(self):\n # Open a memray tracker per worker process, if requested. Each\n # worker writes its own .bin file so traces don't collide.\n out_dir = os.environ.get(_MEMRAY_OUT_DIR_ENV)\n if out_dir:\n pathlib.Path(out_dir).mkdir(parents=True, exist_ok=True)\n bin_path = pathlib.Path(out_dir) / (\n f\"memray-{os.getpid()}-{uuid.uuid4().hex}.bin\"\n )\n self._tracker = memray.Tracker(\n str(bin_path), native_traces=False, follow_fork=False\n )\n self._tracker.__enter__()\n self.model = load_model()\n\n def __call__(self, text: str) -> list[float]:\n if self.model is None:\n self.setup()\n return self.model.embed(text)\n";
|
|
|
|
export const PyRayClusterProfile = "from geneva.runners.ray._mgr import ray_cluster\n\nwith ray_cluster(\n local=True,\n extra_env={\"MY_UDF_MEMRAY_OUT_DIR\": \"/tmp/my-udf-profile\"},\n):\n table.backfill(\"embedding\", concurrency=1)\n";
|
|
|
|
export const PyStatefulUdfClass = "@geneva.udf(data_type=pa.list_(pa.float32(), 512))\nclass MyEmbedding:\n def __init__(self):\n self.model = None\n\n def setup(self):\n self.model = load_model() # allocated once per actor\n\n def __call__(self, text: str) -> list[float]:\n if self.model is None:\n self.setup()\n return self.model.embed(text)\n";
|
|
|
|
export const PyTorchInferenceMode = "def __call__(self, text: str) -> list[float]:\n with torch.inference_mode(): # <-- prevents autograd graph retention\n return self.model.encode(text)\n";
|
|
|