Handle plugins that crash during load #1280
Description
I installed a plugin that turned out to crash on my machine:
> llm install llm-gpt4all
...
> llm --help
Traceback (most recent call last):
File "/Users/jcushman/.local/bin/llm", line 4, in <module>
from llm.cli import cli
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/llm/cli.py", line 3761, in <module>
load_plugins()
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/llm/plugins.py", line 28, in load_plugins
pm.load_setuptools_entrypoints("llm")
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/pluggy/_manager.py", line 416, in load_setuptools_entrypoints
plugin = ep.load()
^^^^^^^^^
File "/Users/jcushman/.local/share/uv/python/cpython-3.12.5-macos-x86_64-none/lib/python3.12/importlib/metadata/__init__.py", line 205, in load
module = import_module(match.group('module'))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/jcushman/.local/share/uv/python/cpython-3.12.5-macos-x86_64-none/lib/python3.12/importlib/__init__.py", line 90, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/llm_gpt4all.py", line 1, in <module>
from gpt4all import GPT4All as _GPT4All
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/gpt4all/__init__.py", line 1, in <module>
from .gpt4all import CancellationError as CancellationError, Embed4All as Embed4All, GPT4All as GPT4All
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/gpt4all/gpt4all.py", line 23, in <module>
from ._pyllmodel import (CancellationError as CancellationError, EmbCancelCallbackType, EmbedResult as EmbedResult,
File "/Users/jcushman/.local/share/uv/tools/llm/lib/python3.12/site-packages/gpt4all/_pyllmodel.py", line 37, in <module>
raise RuntimeError(textwrap.dedent("""\
RuntimeError: Running GPT4All under Rosetta is not supported due to CPU feature requirements.
Please install GPT4All in an environment that uses a native ARM64 Python interpreter.
It took me longer than it should have to get back to a stable state, since llm uninstall couldn't run — one correct answer I could have noticed earlier is to copy "/Users/jcushman/.local/share/uv/tools/llm/" from the stack trace and run /Users/jcushman/.local/share/uv/tools/llm/bin/python -m pip uninstall llm-gpt4all.
I wonder if llm/plugins.py could catch plugin errors by default. (Maybe with an environment variable or something to not catch during plugin development, if that seems useful.)
The Installing Plugins docs could also have a section on uninstalling, with and without llm install -- maybe not necessary if errors are caught.