This is a great showcase and argument for advocating to re-host the origin-destination data as parquet.
Meanwhile, I tested a few data repositories, such as Zenodo, Harvard Dataverse, OSF, Figshare. None provide a good way to access a parquet file. One has to use their API and download the whole file to work with it. Does not seem to be possible to query them.
Very interesting, seems like Amazon S3 is a bit of a standard, I wonder how much hosting would cost. We could add some mechanism to limit the rate e.g. API key/password.
https://observablehq.com/@fil/insee-parquet
This is a great showcase and argument for advocating to re-host the origin-destination data as parquet.
Meanwhile, I tested a few data repositories, such as Zenodo, Harvard Dataverse, OSF, Figshare. None provide a good way to access a parquet file. One has to use their API and download the whole file to work with it. Does not seem to be possible to query them.