wehub-resource-sync
30 lines
1.9 KiB
Markdown
30 lines
1.9 KiB
Markdown
# Custom collection adapter
|
|
|
|
Especially when dealing with large amounts of data, it can be both slow and inconvenient to convert
|
|
your data into the components & datatypes provided by the Rerun SDK in order to log it.
|
|
|
|
This example demonstrates how to solve this using [`rerun::ComponentAdapter`](https://ref.rerun.io/docs/cpp/stable/structrerun_1_1CollectionAdapter.html) for your own types:
|
|
Whenever you have data that is continuous in memory and binary compatible with an existing Rerun component,
|
|
you can adapt it to map to the respective Rerun component.
|
|
For non-temporary objects that live until [`rerun::RecordingStream::log`](https://ref.rerun.io/docs/cpp/stable/classrerun_1_1RecordingStream.html#af7a14a7e2c3029ef1679ff9fd680129d) returns,
|
|
it is typically safe to do so with a zero-copy "borrow".
|
|
This means that in those cases [`rerun::Collection`](https://ref.rerun.io/docs/cpp/stable/classrerun_1_1Collection.html) will merely store a pointer and a length to your data.
|
|
|
|
While collection adapters are primarily used with components, they are also useful for all other usages of
|
|
Rerun's collection type. E.g. the backing buffer of [`rerun::TensorData`](https://ref.rerun.io/docs/cpp/stable/structrerun_1_1datatypes_1_1TensorBuffer.html)
|
|
is also a [`rerun::Collection`](https://ref.rerun.io/docs/cpp/stable/classrerun_1_1Collection.html)
|
|
allowing you to ingest large amounts of data without a copy and the convenience custom adapters can provide.
|
|
|
|
|
|
To build it from a checkout of the repository (requires a Rust toolchain):
|
|
```bash
|
|
cmake .
|
|
cmake --build . --target example_custom_collection_adapter
|
|
./examples/cpp/minimal/example_custom_collection_adapter
|
|
```
|
|
|
|
---
|
|
|
|
* 🚧 TODO(#4257): In the future, adapters will be able to provide simple data transformations like strides to be done as a borrow.
|
|
* 🚧 TODO(#3977): We plan to provide adapters for common types from Eigen and OpenCV.
|