The new describe is awesome, but if I put in non-compliant geoparquet there's little messaging that I have a file that's not quite right:
gpq describe taxi.parquet
╭────────────┬────────┬─────────────────────────────────┬────────────┬─────────────╮
│ COLUMN │ TYPE │ ANNOTATION │ REPETITION │ COMPRESSION │
├────────────┼────────┼─────────────────────────────────┼────────────┼─────────────┤
│ OBJECTID │ int32 │ int(bitwidth=32, issigned=true) │ 0..1 │ snappy │
│ Shape_Leng │ double │ │ 0..1 │ snappy │
│ Shape_Area │ double │ │ 0..1 │ snappy │
│ zone │ binary │ string │ 0..1 │ snappy │
│ LocationID │ int32 │ int(bitwidth=32, issigned=true) │ 0..1 │ snappy │
│ borough │ binary │ string │ 0..1 │ snappy │
│ geom │ binary │ │ 0..1 │ snappy │
├────────────┼────────┼─────────────────────────────────┴────────────┴─────────────┤
│ ROWS │ 262 │ │
╰────────────┴────────┴────────────────────────────────────────────────────────────╯
If I convert it then I get an additional row:
├──────────┼────────┼────────────┴────────────┴─────────────┴──────────┴────────────────┴────────┴────────┤
│ ROWS │ 3233 │ │
│ VERSION │ 1.0.0 │ │
╰──────────┴────────┴─────────────────────────────────────────────────────────────────────────────────────╯
I think it'd be nice to try to always put something about the compliance. Like maybe always have VERSION, but if it's not compliant than say non-compliant (might also be nice to call it 'geoparquet version' or something). It could also be nice to say if it's a 'compatible parquet' file, like it's geom and data looks like 4326, and recommend people use gpq convert.
The new describe is awesome, but if I put in non-compliant geoparquet there's little messaging that I have a file that's not quite right:
If I convert it then I get an additional row:
I think it'd be nice to try to always put something about the compliance. Like maybe always have VERSION, but if it's not compliant than say
non-compliant(might also be nice to call it 'geoparquet version' or something). It could also be nice to say if it's a 'compatible parquet' file, like it's geom and data looks like 4326, and recommend people use gpq convert.