feat(loader): .gpkg loader

[jfb-h]

Reopen #14, this time with tests!

[tecosaur]

Love it, thanks for adding tests too 😍

[tecosaur]

You don't need to using ArchGDAL in runtests.jl, it will be automatically loaded as normal.

[tecosaur]

I can take care of that though if I merge this locally (might make this PR show up as "closed" though, but that's just cosmetics at the end of the day, the commit log will still show this contribution as coming from you).

[jfb-h]

You don't need to using ArchGDAL in runtests.jl, it will be automatically loaded as normal.

I thought so, but when I don't explicitly load it, the test fails with a TransformerError: Data set "eurostat-gpkg" could not be loaded in any form.. Did I miss something?

[tecosaur]

🤔 I'll have a look locally

[jfb-h]

Hi @tecosaur, I know you have a lot on your hands, but is there any chance this could be merged soonish? I have a project where I'd love to transition to DataToolkit.jl but which requires handling .gpkg files. Thanks a lot for all your effort!

[tecosaur]

Sorry for the delay (I'm beginning to get a bit busier, feel free to ping me like this though 🙂), and thanks for the contribution.

I'll give this a second look now.

[tecosaur]

I've just made a few tweaks, and realised the what the problem was/reason why ArchGDAL had to be loaded explicitly. It's because the ArchGDAL type specified in supportedtypes doesn't exist before the package is loaded, and so DataToolkitBase can't infer anything about the type. I've thought of an only slightly kludgy fix, I'll push that and merge this in a sec.

[tecosaur]

The failing CI check now seems to be

gpkg: Test Failed at /home/tec/.julia/dev/DataToolkitCommon/test/runtests.jl:28
  Expression: ArchGDAL.getlayer(geo, 0) |> length == 16
   Evaluated: 1025 == 16

let me know if this represents an actual issue, or if this should just be ... == 1025.

[jfb-h]

Ah damn, I swear I tested that locally... Yeah it should indeed be 1025, I must have mixed up the files. Thanks a lot for looking into this!