Add FlatGeobuf encoding #32
Conversation
|
|
||
| ## Potential issues in conversion | ||
|
|
||
| TBD |
There was a problem hiding this comment.
fiboa allowes enums for strings and integers, but flatbuffers don't.
It think we should mention this here.
https://flatbuffers.dev/flatbuffers_guide_writing_schema.html
| | binary | Binary | | ||
| | string<br />charset: UTF-8 | String | | ||
| | array | Json Array | | ||
| | object<br />keys: string<br />values: any | Json Object | |
There was a problem hiding this comment.
What about structs? I think both could work depending on the specificity of the schema.
I think fiboa, also foe GeoParquet, should do a better job at explaining which schema leads to a struct and which to a map. I'll open an issue: #34
| | date | DateTime (ISO 8601 date) | | ||
| | date-time<br />with milliseconds<br />timezone: UTC | DateTime (ISO 8601 date time) | |
There was a problem hiding this comment.
I couldn't find datetime data types in the specification. Can you help me finding it? Otherwise it should probably string with a pre-defined format (e.g. ISO 8601 / RFC 3339)
There was a problem hiding this comment.
Yes, I was refering to this line: https://github.com/flatgeobuf/flatgeobuf/blob/2828e07fd759aa8b44c6b1aaab41320b22dd0d29/src/fbs/header.fbs#L40 It is (probably) an ISO 8601 string, but it is it's own ColumnType in the FlatGeobuf schema.
Related comment: #32 (comment)
There was a problem hiding this comment.
Thanks, that clairifies it. We need to clarify how we handle date only more practically. What happens with the missing components (hour, minute, ...). Maybe add a link somewhere to the allowed types.
There was a problem hiding this comment.
Yes, I agree DateTime is not well defined. I put a link to the allowed column types here: https://github.com/fiboa/specification/pull/32/files#diff-c33cc7b35504997d9be5d9db7ecaf8f5975eef4b5fe303b28c915033bfba02adR4 But we can also link to https://unpkg.com/[email protected]/dist/doc/enums/generic.ColumnType.html if you prefer.
There was a problem hiding this comment.
Didn't see that link before, looks good. So here the Date issue is still open.
Co-authored-by: Matthias Mohr <[email protected]>
|
Thanks for the contribution! This is awesome to see, and I love flatgeobuf. My only hesitation is if there's going to be more of a burden to support 3 (or more, as geopackage would come next) 'officially' as part of fiboa. I had thought we'd welcome mappings to flatgeobuf and others, but mark things in some way to say that the mappings are more 'informative' and we're not committing to having all the tooling work with alternate encodings. |
Related Issue(s): #24
PR Checklist:
or a CHANGELOG entry is not required.