Remove pyarrow upperbound version constraint

[kevinjqliu]

Changes included in this PR

Allows upstream library, such as pyiceberg, to freely use newer versions of pyarrow

Testing strategy

User facing changes

Checklist

• Pipelines passed before requesting review. To run CI you must include [run CI] in your commit message.
• I am familiar with the Contributing Guide
• I have installed + ran pre-commit hooks.

[kevinjqliu]

cc @ehsantn could you take a look at this change? I'm trying to use pyiceberg with pyarrow 21

[ehsantn]

Thanks for opening this @kevinjqliu. The challenge is that Bodo uses Arrow C++, which makes building pip wheels difficult for different Arrow versions. We currently have to pin a version, unless if we statically bundle Arrow as well. I'll explore this with the team.

The upcoming PyIceberg 0.10 release enables upgrading to Arrow 21, which we will do to unblock ASAP.

Other thoughts @scott-routledge2 @IsaacWarren ?

[klopppo]

I'm getting error while importing bobo
AttributeError: module 'pyarrow.lib' has no attribute 'ListViewType'

I just installed with pip install -u bobo

my yarrow: (from bodo) (19.0.1)

is there another version I should install? please let me know

[scott-routledge2]

I'm getting error while importing bobo AttributeError: module 'pyarrow.lib' has no attribute 'ListViewType'

I just installed with pip install -u bobo

my yarrow: (from bodo) (19.0.1)

is there another version I should install? please let me know

@klopppo Your pyarrow version sounds correct to me. If you let me know what machine type/python version you are using as well as any additional packages you have in your environment I could look into it for you. Also you could try installing with conda in a separate environment if you have that (conda install -c conda-forge bodo) as sometimes that can be a bit more robust than pip.