docs: Update docs with many of the recent changes

README.md

@@ -28,7 +28,7 @@ evalio can be used both as a python library and as a CLI for both datasets and p
 Once evalio is installed, datasets can be listed and downloaded via the CLI interface. For example, to list all datasets and then download a sequence from the hilti-2022 dataset,
 ```bash
 evalio ls datasets
-evalio download hilti_2022/basement_2
+evalio dl hilti_2022/basement_2
 ```
 
 Once downloaded, a trajectory can then be easily used in python,
@@ -108,4 +108,4 @@ If you use evalio in your research, please cite the following paper,
       primaryClass={cs.RO},
       url={https://arxiv.org/abs/2507.16000}, 
 }
-```
+```

docs/examples/data_loading.md

@@ -81,12 +81,12 @@ from evalio.rerun import convert
 
 # Initialize rerun
 rr.init("evalio")
-rr.connect_tcp()
+rr.connect_grpc()
 
 # Stream lidar scans to rerun
 for scan in Hilti2022.basement_2.lidar():
     rr.set_time("timeline", timestamp=scan.stamp.to_sec())
     rr.log("lidar", convert(scan, color="z"))
 ```
 
-If anything is unclear, please open an issue on the evalio repository. We are always looking to improve the documentation and make it easier to use.
+If anything is unclear, please open an issue on the evalio repository. We are always looking to improve the documentation and make it easier to use.

docs/examples/dataset.md

@@ -70,7 +70,7 @@ Additionally, there is a number of optional methods that you can implement to ad
 
     def vehicle(self) -> str: ...
 
-    def download(self) -> str: ...
+    def download(self) -> None: ...
 
     def quick_len(self) -> Optional[int]: ...
 ```
@@ -82,4 +82,4 @@ The next three are again self-explanatory, all of which provide information for
 
 `quick_len` returns a hardcoded number of scans in a dataset, used for `evalio ls` and for computing time estimates in `evalio run`. If not set, evalio will load the data to compute the length.
 
-That's all there is to it! Datasets are fairly simple - mostly just parameter setting and easy-to-use iterator wrappers. If you have an dataset implementation of an open-source dataset, feel free to make a PR to add it to evalio so others can use it as well.
+That's all there is to it! Datasets are fairly simple - mostly just parameter setting and easy-to-use iterator wrappers. If you have an dataset implementation of an open-source dataset, feel free to make a PR to add it to evalio so others can use it as well.

docs/examples/evaluation.md

@@ -63,6 +63,6 @@ Results will be saved to the `output_dir` specified in the config file, with nes
 
 Once trajectories have been run, statistics can be calculated,
 ```bash
-evalio stats -d results -m mean -w 200 -s RTEt
+evalio stats results --metric mean --w-meters 200 --sort RTEt_200.0m
 ```
-With `-m/--metric` specifying the metric to calculate with options including mean, median, and sse and `-w/--window` specifying the window size for RTE, with a default of 100 scans. Only first part of all trajectories can also be done using the `-l/--length` option. Sorting of the results can be done with the `-s/--sort` option, with any column heading being an allowed option. 
+With `--metric` specifying the metric to calculate with options including mean, median, and sse, and `--w-meters` / `--w-seconds` specifying RTE windows (defaults to `--w-meters 30` and can be repeated). Only first part of all trajectories can also be done using the `-l/--length` option. Sorting of the results can be done with `-S/--sort`, with any column heading being an allowed option.

docs/examples/odometry.md

@@ -44,12 +44,12 @@ from evalio.types import Trajectory
 from evalio.datasets import NewerCollege2020
 from evalio import stats
 
-traj = Trajectory.from_experiment("odometry.csv")
+traj = Trajectory.from_file("odometry.csv")
 gt = NewerCollege2020.short_experiment.ground_truth()
 
 # Align the odometry to the ground truth
 traj_aligned, gt_aligned = stats.align(traj, gt)
 
 # Compute metrics as desired (will align if not already aligned)
 error =  stats.rte(traj, gt).mean()
-```
+```

docs/examples/pipeline.md

@@ -33,7 +33,7 @@ To create a pipeline, simply inherit from the `Pipeline` class,
 
         # Getters
         def pose(self) -> SE3: ...
-        def map(self) -> list[Point]: ...
+        def map(self) -> dict[str, list[Point]]: ...
 
         # Setters
         def set_imu_params(self, params: ImuParams): ...
@@ -44,7 +44,7 @@ To create a pipeline, simply inherit from the `Pipeline` class,
         # Doers
         def initialize(self): ...
         def add_imu(self, mm: ImuMeasurement): ...
-        def add_lidar(self, mm: LidarMeasurement) -> list[Point]: ...
+        def add_lidar(self, mm: LidarMeasurement) -> None: ...
     ```
 
 === "C++"
@@ -72,7 +72,7 @@ To create a pipeline, simply inherit from the `Pipeline` class,
 
         // Getters
         const SE3 pose() { ... };
-        const std::vector<Point> map() { ... };
+        const evalio::Map<> map() { ... };
 
         // Setters
         void set_imu_params(ImuParams params) { ... };
@@ -83,7 +83,7 @@ To create a pipeline, simply inherit from the `Pipeline` class,
         // Doers
         void initialize() { ... };
         void add_imu(ImuMeasurement mm) { ... };
-        std::vector<Point> add_lidar(LidarMeasurement mm) { ... };
+        void add_lidar(LidarMeasurement mm) { ... };
     }
     ```
 
@@ -93,6 +93,8 @@ We'll cover each section of methods in turn.
 
 The first four methods are all static methods that provide information about the pipeline. `version`, `url`, and `name` are all self-explanatory. `default_params` is a static method that returns a dictionary of the default parameters for the pipeline. This is used to verify parameters before they are passed in, as well as ensure a consistent output for each run.
 
+In C++ there is additionally a number of helper type conversion functions that can make converting between iterators, point types, and geometry types simpler. A good example of this can be found in the `lio_sam.h` binding where it is used to convert pose and point types. These converters can additionally be leveraged by the `save` methods described below.
+
 ## Getters
 
 The next two methods are getters for the pose and map. The pose is the most up-to-date estimate for the IMU and is polled after each lidar measurement is passed in. 
@@ -112,7 +114,11 @@ Arguably the most important part.
 
 `add_imu` is called for each IMU measurement. This is where the IMU data is processed and used to update the pose.
 
-`add_lidar` is called for each lidar measurement. This is where the lidar data is processed and used to update the map. It returns a list of features were extracted from the scan and are used for visualization.
+`add_lidar` is called for each lidar measurement. This is where the lidar data is processed and used to update the map. 
+
+Saving poses can be done asynchronously, using `save(stamp, pose)`. In C++, the type of `pose` can be anything that has the method `evalio::convert<evalio::SE3>(const MyPose& pose)` implemented.
+
+For visualization, features can be save similarly with either `save(stamp, {"key": features})` in python, or `save(stamp, "key1", feat1, "key2", feat2)` in C++. Again, `feat1` and `feat2` can be any type that are iterators with their internal point types convertible to `evalio::Point`. 
 
 ## C++ Building
 
@@ -125,11 +131,11 @@ NB_MODULE(_core, m) {
 
   // Only have to override the static methods here
   // All the others will be automatically inherited from the base class
-  nb::class_<MyCppPipeline, evalio::Pipeline>(m, "MyCppPipeline")
+  nb::class_<MyPipeline, evalio::Pipeline>(m, "MyPipeline")
       .def(nb::init<>())
-      .def_static("name", &MyCppPipeline::name)
-      .def_static("url", &MyCppPipeline::url)
-      .def_static("default_params", &MyCppPipeline::default_params);
+      .def_static("name", &MyPipeline::name)
+      .def_static("url", &MyPipeline::url)
+      .def_static("default_params", &MyPipeline::default_params);
 }
 ```
 
@@ -145,4 +151,4 @@ We recommend then setting everything up to be built with [`scikit-build-core`](h
     m.def("abi_tag", []() { return nb::detail::abi_tag(); });
     ```
 
-That's all there is to it! Pipelines should be fairly easy to implement and are usually just a simple wrapper around your existing code to provide a common interface. Once your pipeline is open-source/published/etc, feel free to make a PR to add it to evalio. This both improves the visibility of your work and of evalio.
+That's all there is to it! Pipelines should be fairly easy to implement and are usually just a simple wrapper around your existing code to provide a common interface. Once your pipeline is open-source/published/etc, feel free to make a PR to add it to evalio. This both improves the visibility of your work and of evalio.

docs/install.md

@@ -23,7 +23,7 @@ make
 ```
 
 ### Pipelines
-By default, pipelines are not included due to their large dependencies. We use vpckg to handle a reliable build of these dependencies and pipelines. vcpkg and the pipelines can be setup via running
+By default, pipelines are not included due to their large dependencies. We use vcpkg to handle a reliable build of these dependencies and pipelines. vcpkg and the pipelines can be setup via running
 ```bash
 ./cpp/setup_pipelines.sh
 ```

docs/quickstart.md

@@ -12,7 +12,7 @@ evalio can be used both as a python library and as a [CLI](ref/cli.md) for both
 Once evalio is installed, datasets can be listed and downloaded via the [CLI](ref/cli.md) interface. For example, to list all datasets and then download a sequence from the hilti-2022 dataset,
 ```bash
 evalio ls datasets
-evalio download hilti_2022/basement_2
+evalio dl hilti_2022/basement_2
 ```
 evalio downloads data to the path given by `-D`, `EVALIO_DATA` environment variable, or if both are unset to the local folder `./evalio_data`. All the trajectories in a dataset can also be downloaded by using the wildcard `hilti_2022/*`, making sure to escape the asterisk as needed.
 
@@ -67,10 +67,10 @@ import rerun as rr
 from evalio.rerun import convert
 
 rr.init("evalio")
-rr.connect_tcp()
+rr.connect_grpc()
 for scan in ds.Hilti2022.basement_2.lidar():
     rr.set_time("timeline", timestamp=scan.stamp.to_sec())
-    rr.log("lidar", convert(scan, color=[255, 0, 255]))
+    rr.log("lidar", convert(scan, color=(255, 0, 255)))
 ```
 
 !!! note
@@ -143,4 +143,4 @@ where m -> map, s -> scan, i -> intensity image, and f -> extracted features can
 
 That's about the gist of it! Try playing around the [CLI](ref/cli.md) interface to see what else is possible. Feel free to open an issue if you have any questions, suggestions, or problems. 
 
-Additionally, we recommend checking out the examples section for specific use cases for evalio.
+Additionally, we recommend checking out the examples section for specific use cases for evalio.

docs/ref/pipelines.md

@@ -5,12 +5,13 @@ For more information about the pipelines included in evalio, see the [included p
         members:
             - Pipeline
             - CTICP
-            - KissICP
+            - DLIO
+            - FORM
             - GenZICP
+            - KissICP
             - LOAM
             - LioSAM
             - MadICP
-            - FORM
             - PipelineNotFound
             - UnusedPipelineParam
             - InvalidPipelineParamType

python/evalio/cli/__init__.py

@@ -55,9 +55,6 @@ def module_callback(value: Optional[list[str]]) -> list[Any]:
 
 @app.callback()
 def global_options(
-    # Marking this as a str for now to get autocomplete to work,
-    # Once this fix is released (hasn't been as of 0.15.2), we can change it to a Path
-    # https://github.com/fastapi/typer/pull/1138
     data_dir: Annotated[
         Optional[Path],
         typer.Option(

python/evalio/cli/stats.py

@@ -286,7 +286,7 @@ def evaluate_typer(
     # Parse some of the options
     if only_complete and only_failed:
         raise typer.BadParameter(
-            "Can only use one of --only-complete, --only-incomplete, or --only-failed."
+            "Can only use one of --only-complete or --only-failed."
         )
 
     # Parse the filtering options

python/evalio/datasets/base.py

@@ -233,7 +233,7 @@ def _fail_not_downloaded(self):
         if not self.is_downloaded():
             # TODO: Make this print with rich?
             raise ValueError(
-                f"Data for {self} not found, please use `evalio download {self}` to download"
+                f"Data for {self} not found, please use `evalio dl {self}` to download"
             )
 
     @classmethod