thinkalexandria / diesel_geometry Goto Github PK

First, congratulations for this project! I've working for some time with diesel and postgres for radial distance computations using Point and the earthdistance module, but I hadn't time to isolate the API and publish a crate yet.

My question is if I should try to merge my work with earthdistance (or something that uses extensions, such as geom os gis) on this crate or try to publish it elsewhere using only the Point data type.

Thanks,

I know there are a few good examples as tests, but it would be good if they were available on the README or the first page of the docs.

The usage is pretty straightforward besides having to include the Point type on the table! macro. I had some hard time having to figure out this on the first time working with custom diesel types.

Diesel now uses a diesel.toml file that allows configuring the schema generation process:
http://diesel.rs/guides/configuring-diesel-cli/
It also allows adding use statements:

[print_schema]
# Add types from `diesel_full_text_search` like `tsvector`
import_types = ["diesel::sql_types::*", "diesel_full_text_search::types::*"]

Would it be possible to support diesel_geometry types with automatic schema generation this way?
Or would diesel_cli also have to be aware of the types in this crate (Point etc.) in order to generate schemas with such columns?

I'm looking forward to using this crate with PostGIS.
Do you have plans to support the other geometry types like circle? :)

I tried to use it with geography point type geography(point, 4326) not null in postgres and PgPoint in my model struct, but it's not deserializing the values correctly:

When I insert

ST_SetSRID(ST_MakePoint(13.409779,52.520645), 4326)::geography

and then print it in my Rust executable, it prints:

PgPoint(0.0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000007746818039357731, 0.00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000013770074297002852)

:/

Please add optional support for serde's Serialize/Deserialize traits.
Like this:

serde = { version = "1.0", features = ["derive"], optional = true }

#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub struct PgPoint(pub f64, pub f64);

Thanks!