Vikash

Setting maxH in the box element causes the dropdown position to change (it renders over the input element). I want the dropdown to still be below the input.

Yes, I was thinking about portals as well. But I was playing around with your code and I am not sure how to fix the position to be directly below...

@ritchie46 how would you do horizontal concatenation using a join when there are no common columns? It would be good to have a horizontal concat and hstack working for lazyframes.

@zundertj I just ran ```python import polars as pl import numpy as np df = pl.DataFrame({ 'doc_id': np.random.randint(0, high=1000000, size=1000000000), 'passage_id': np.random.randint(0, high=100, size=1000000000), 'score': np.random.random(size=1000000000) }).with_column(pl.lit('40240gh32152n').alias('memo_id')) df.write_parquet('test.bin', row_group_size=1000000) ```...

@ritchie46 Tried again with `df.write_parquet('test.bin', row_group_size=1024)`. Somehow this runs out of memory too, but even slower than `df.write_parquet('test.bin', row_group_size=1000000)`. Maybe the memory is not being freed after each row group...

@ritchie46 Nope, still getting the same problem with 0.14.18. It ran out of memory after about 2 hours and it still didn't manage to finish writing the parquet.

@ritchie46 Is there anything else I can help with reproducing the issue? The fact that it only happens for parquet and not csv is strange. Also it doesn't seem to...

@ritchie46 I am testing the code again with row group size of 1 million, but this time with a memory profiler to track memory usage over time. Also I have...

Doing the same thing for CSV: ```python import polars as pl import numpy as np from memory_profiler import memory_usage def ram_test(): df = pl.DataFrame({ 'doc_id': np.random.randint(0, high=1000000, size=1000000000), 'passage_id': np.random.randint(0,...

I think the authorization issues should probably be fixed once https://github.com/python-poetry/poetry/pull/5442 is merged