[BUG] Sequences of fixed vs varying size handled inconsistently

🐛🐛 Bug Report

⚗️ Current Behavior

Input Code

import hub
from hub.schema import Text, Sequence

@hub.transform(schema={"txts": Sequence(dtype=Text(max_shape=(100,)))})
def transform_ds(txts):
    return {"txts": txts}

# This works
rows = [
    ["one", "two"],
    ["one", "two", "three"],
]
transform_ds(rows).store("./tmp_ds")

# This throws `ValueError: parameter 'value': expected array with shape (2,), got (2, 2)`
rows = [
    ["one", "two"],
    ["one", "two"],
]
transform_ds(rows).store("./tmp_ds")

Expected behavior

I would expect it to work consistently across “ragged” sequences (rows with varying length) as well as ones where each row has the same number of elements.

⚙️ Environment

• Python version: 3.6.9
• Packages: [ hub==1.3.5, zarr==2.7.1]

🧰 Possible Solution (optional)

zarr.indexing.BasicIndexer.shape == (2,) in both of the above cases, which seems wrong. Not sure if this is a bug with zarr or hub is calling zarr incorrectly.

[dhiganthrao]

Closing this issue, as dynamic sequences are properly handled in 2.0. @mattiasarro, @FayazRahman - I’m also excited to announce we’ve released the 2.0 community beta - you can try it out using pip (pip install hub). You can access the docs here, and access some small tutorials to get started with 2.0 here! You can also let me know if you run into any issues, I’ll be happy to help! If you haven’t, please consider joining our Community Slack to receive further updates.

[dhiganthrao]

Hey there! Thanks for bringing this issue up. We’ll try reproducing this issue and get back to you in some time!