🔥 Remove cog3pio backend engine, drop xarray dependency (#89)

* ➖ Drop xarray dependency, remove cog3pio backend engine

Deciding to remove the image-tiff CPU-based Cog3pioBackendEntrypoint, in favour of the nvtiff CUDA-based backend happening over at cupy-xarray. Remove all code related to xarray, and also the rioxarray comparison benchmark. Users should look into other CPU-based TIFF readers (e.g. async-tiff) instead. This will also trim our dependency footprint considerably for easier maintenance and clearer focus on DLPack.

* 📝 Delete or cross-out xarray related documentation

Remove xarray section from quickstart and API pages, deleted mention of xarray in recent changelog, and crossed it out in main README.md.

* 📝 Update roadmap to take include write support and SIMD wishlist

Take out async-tiff refactor since async-geotiff is out. Want to double-down more on hardware-accelerated decoding instead.

Author: weiji14

.release-plz.toml

@@ -61,6 +61,7 @@ commit_preprocessors = [
     { pattern = ':boom:', replace = "💥" },
     { pattern = ':bug:', replace = "🐛" },
     { pattern = ':construction_worker:', replace = "👷" },
+    { pattern = ':fire:', replace = "🔥" },
     { pattern = ':heavy_minus_sign:', replace = "➖" },
     { pattern = ':heavy_plus_sign:', replace = "➕" },
     { pattern = ':label:', replace = "🏷️" },
@@ -85,7 +86,7 @@ commit_parsers = [
     { message = "^(💥|:boom:|🚀|:rocket:)", group = "<!-- 0 --> 🌈 Highlights" },
     { message = "^(✨|:sparkles:)", group = "<!-- 1 --> ✨ Features" },
     { message = "^(🐛|:bug:)", group = "<!-- 2 --> 🐛 Bug Fixes" },
-    { message = "^(♻️|:recycle:|🚚|:truck:|🎨|:art:)", group = "<!-- 3 --> 🏭 Refactors" },
+    { message = "^(♻️|:recycle:|🚚|:truck:|🎨|:art:|🔥|:fire:)", group = "<!-- 3 --> 🏭 Refactors" },
     { message = "^(📝|:memo:|🔍️|:mag:|🏷️|:label:)", group = "<!-- 4 --> 📝 Documentation" },
     { message = "^(👷|:construction_worker:|🔧|:wrench:|⬆️|:arrow_up:|➕|:heavy_plus_sign:|➖|:heavy_minus_sign:|⬇️|:arrow_down:|📦️|:package:|📌|:pushpin:|🚩|:triangular_flag_on_post:|🔒️|:lock:|🚨|:rotating_light:|🌱|:seedling:|🔊|:loud_sound:)", group = "<!-- 5 --> 🧰 Maintenance" },
 ]

README.md

@@ -20,8 +20,8 @@ Yet another attempt at creating a GeoTIFF reader, in Rust, with Python bindings.
       [`object-store`](https://crates.io/crates/object_store))
 
 2024 Q2-Q4:
-- [x] Integration with `xarray` as a
-      [`BackendEntrypoint`](https://docs.xarray.dev/en/v2024.02.0/internals/how-to-add-new-backend.html)
+- [x] ~Integration with `xarray` as a
+      [`BackendEntrypoint`](https://docs.xarray.dev/en/v2024.02.0/internals/how-to-add-new-backend.html)~ (removed)
 - [x] Implement single-band GeoTIFF reader for multiple dtypes (uint/int/float) (based
       on [`geotiff`](https://crates.io/crates/geotiff) crate)
 
@@ -34,9 +34,11 @@ Yet another attempt at creating a GeoTIFF reader, in Rust, with Python bindings.
 - [x] GPU-based decoding (via [`nvTIFF`](https://crates.io/crates/nvtiff-sys))
 
 2026:
-- [ ] Asynchronous I/O (refactor to [`async-tiff`](https://crates.io/crates/async-tiff))
 - [ ] Direct-to-GPU loading
+- [ ] Write support
 
+2027:
+- [ ] SIMD decoding
 
 ## Related crates
 

docs/api.md

@@ -8,14 +8,6 @@ These are the Python docs for `cog3pio`, for the Rust docs, see
 ::: cog3pio.CudaCogReader
 ::: cog3pio.CogReader
 
-
-## Xarray
-
-::: cog3pio.xarray_backend.Cog3pioBackendEntrypoint
-    options:
-      members:
-      - open_dataset
-
 ## NumPy
 
 ::: cog3pio.read_geotiff

docs/changelog.md

@@ -9,7 +9,7 @@ All notable changes to this project will be documented in this file.
 ### <!-- 0 --> 🌈 Highlights
 
 - 💥 Read into CUDA memory DLPack tensor ([#57](https://github.com/weiji14/cog3pio/pull/57))
-- ⬆️ SPEC 0: Bump min version to Python 3.13, xarray 2024.10.0 ([#66](https://github.com/weiji14/cog3pio/pull/66))
+- ⬆️ SPEC 0: Bump min version to Python 3.13 ([#66](https://github.com/weiji14/cog3pio/pull/66))
 
 ### <!-- 1 --> ✨ Features
 

docs/quickstart.md

@@ -1,18 +1,16 @@
 # Quickstart
 
-There are three ways to read a GeoTIFF with `cog3pio`'s Python bindings into CPU memory.
+There are two ways to read a GeoTIFF with `cog3pio`'s Python bindings into CPU memory.
 Take your pick:
 
-|  Output   | DLPack protocol | coordinates | any dtype |
-|:---------:|:---------------:|:-----------:|:---------:|
-| PyCapsule |      ✅         |     ❌      |     ✅    |
-| xarray    |      ❓         |     ✅      |     ✅    |
-| numpy     |      ✅         |     ❌      |     ❌    |
+|  Output   | DLPack protocol | any dtype |
+|:---------:|:---------------:|:---------:|
+| PyCapsule |      ✅         |      ✅    |
+| numpy     |      ✅         |      ❌    |
 
 Notes:
 
 - [DLPack - an in-memory tensor structure](https://data-apis.org/array-api/latest/design_topics/data_interchange.html#dlpack-an-in-memory-tensor-structure)
-- Coordinates in xarray are computed from the GeoTIFF's affine transformation
 - Currently supported dtypes include uint (u8/u16/u32/u64), int (i8/i16/i32/i64) and
   float (f16/f32/f64).
 
@@ -44,24 +42,6 @@ assert array.dtype == "float16"
 # ...
 ```
 
-## Xarray
-
-Read GeoTIFF file from a HTTP url via the
-[`Cog3pioBackendEntrypoint`](api#cog3pio.xarray_backend.Cog3pioBackendEntrypoint) engine
-into an [xarray.DataArray][] object (akin to
-[`rioxarray`](https://corteva.github.io/rioxarray)).
-
-```python
-import xarray as xr
-
-# Read GeoTIFF into an xarray.DataArray
-dataarray: xr.DataArray = xr.open_dataarray(
-    filename_or_obj="https://github.com/cogeotiff/rio-tiler/raw/7.8.0/tests/fixtures/cog_dateline.tif",
-    engine="cog3pio",
-)
-assert dataarray.sizes == {'band': 1, 'y': 2355, 'x': 2325}
-assert dataarray.dtype == "uint16"
-```
 
 ## NumPy
 

pyproject.toml

@@ -23,7 +23,6 @@ classifiers = [
 ]
 dependencies = [
     "numpy>=2.0",
-    "xarray>=2024.10.0",
 ]
 dynamic = ["version"]
 
@@ -32,9 +31,6 @@ dynamic = ["version"]
 "Documentation" = "https://cog3pio.readthedocs.io"
 "Source Code" = "https://github.com/weiji14/cog3pio"
 
-[project.entry-points."xarray.backends"]
-cog3pio = "cog3pio.xarray_backend:Cog3pioBackendEntrypoint"
-
 [project.optional-dependencies]
 cuda = [
     "cupy-cuda13x",
@@ -43,7 +39,6 @@ cuda = [
 [dependency-groups]
 benchmark = [
     "pytest-codspeed",
-    "rioxarray",
 ]
 docs = [
     "mkdocstrings-python",