Parquet loader support ? about nautilus_trader

[cjdsellers]

Lets support parquet, it looks valuable.

I suppose we just need a ParquetTradeDataLoader and ParquetQuoteDataLoader for now. If you're able to provide those with some sample data in the test_kit.data folder with tests then I'm happy to include that.

I think for ease of development and familiarity right now, it'll be a good idea to stick to CSV sample data in general though?

[cjdsellers]

I agree the repo with the sampl data is getting fairly bulky. Definitely I think it's a good idea that any further data be compressed and parquet seems like a good option.

Actually the only reason the rest of the data isn't compressed is due to CI failing issues with pandas which I couldn't get a fix for quickly.

[ian-wazowski]

I think for ease of development and familiarity right now, it'll be a good idea to stick to CSV sample data in general though?

Of course It is good decision, (I love csv.) I just wanted to emphasize that the sample data is getting large that downloading the repository is slow. By compressing it with parquet, I thought I could reduce the total size of the repository.

[jpmediadev]

Actually the only reason the rest of the data isn't compressed is due to CI failing issues with pandas which I couldn't get a fix for quickly.

several times tried to upload gz csv - files were broken

[cjdsellers]

Yes, its a strange error on the gzips.

[cjdsellers]

pyarrow dependency fixed and merged.