Kyle Barron
Kyle Barron
> it's a fairly straightforward piece of metadata to get through other means (e.g., the `st_geometry_column()` generic in sf, or a potential `__geometry__` interface). Can you elaborate on this? >...
> Yes, Pandas does the same thing, although I don't know if they use field metadata or schema metadata. The schema metadata is the problem (field metadata faithfully follows column...
> I think my point is that schema metadata is not a reliable source. Is it any less reliable than the field metadata? We're similarly in a bad position if...
Without this piece of metadata, we necessarily lose information when reading a GeoParquet file into memory with multiple geometry columns, right? Because on the next operation we don't know which...
> If the only method for communicating your output is an `ArrowSchema`, perhaps. I think this is one advantage of pyogrio's reader returning a tuple of `info`, `table`. It seems...
Thanks! I just need to update this PR now that we don't have arrow2 and then I can merge
For some reason the types still aren't loading in some environments, but this is still progress.
As long as you use the Node entry point, that is correct. But if you're using the `/esm` entry point from Node, you still need to call `initWasm`
> does not work as the compiler error says "Cannot find module 'parquet-wasm' or its corresponding type declarations" (tsc 4.9.5). Can you either pin to the most recent stable version...
This should be fixed in 0.6.0, which was just published