Sqlalchemy Turbodbc missing library: libsimdutf.so.9

[amgcc]

After upgrading my conda environment today, package simdutf changed from version 5.4.15 to 5.5.0. Afterward, I started recieving errors with sqlalchemy about a missing library. I'm not sure if this is something that needs to be fixed in the connector, sqlalchemy or turbodbc because I don't understand the issue based on the error message.

    return sa.create_engine(conn_str, **kwargs)
../../../../envs/linux_py311/lib/python3.11/site-packages/sqlalchemy/util/deprecations.py:375: in warned
    return fn(*args, **kwargs)
../../../../envs/linux_py311/lib/python3.11/site-packages/sqlalchemy/engine/create.py:518: in create_engine
    entrypoint = u._get_entrypoint()
../../../../envs/linux_py311/lib/python3.11/site-packages/sqlalchemy/engine/url.py:662: in _get_entrypoint
    cls = registry.load(name)
../../../../envs/linux_py311/lib/python3.11/site-packages/sqlalchemy/util/langhelpers.py:341: in load
    return impl.load()
../../../../envs/linux_py311/lib/python3.11/site-packages/importlib_metadata/__init__.py:209: in load
    module = import_module(match.group('module'))
../../../../envs/linux_py311/lib/python3.11/importlib/__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
../../../../envs/linux_py311/lib/python3.11/site-packages/sqlalchemy_turbodbc/dialect.py:24: in <module>
    from .connector import TurbodbcConnector
../../../../envs/linux_py311/lib/python3.11/site-packages/sqlalchemy_turbodbc/connector.py:16: in <module>
    from turbodbc import make_options
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

    from importlib.metadata import version
    
>   from turbodbc_intern import Megabytes, Rows
E   ImportError: libsimdutf.so.9: cannot open shared object file: No such file or directory

[adamchen]

I don't think this is a sqlalchemy-turbodbc issue, unfortunately! Since 4.11 it seems that when using turbodbc, simdutf needs to be built and available on the host machine. I don't use Conda at all, but this reply over on the turbodbc repo seems to imply that Conda distributes binary versions of turbodbc - perhaps all that is needed for now is to pin your simdutf version so it stays in line with the packaged version? Alternatively, you may have luck compiling the latest simdutf from source if you need the bleeding edge.

I'm going to close this issue as I don't think there's anything to do on the sqlalchemy-turbodbc side, but I'm be happy to reopen if I'm wrong.