We want to lower the threshold for someone to use rerun or re_sdk to just log data to a file, and then open that file in app.rerun.io. That includes minimizing the amount of dependencies one has to build. If someone does cargo add rerun and sees 200 dependencies being pulled in, they may very well abort.
So: to use the SDK to only log to a file should be possible with a minimal set of dependencies; mostly just Arrow.
In the future we could add support for the native viewer streaming and .rrd file that is being written to by the SDK, thus still supporting spawn even without WebSockets:
cargo tree -p rerun -e normal is good tool for working on this, as is cargo build --timings -p rerun --no-default-features
We want to lower the threshold for someone to use
rerunorre_sdkto just log data to a file, and then open that file in app.rerun.io. That includes minimizing the amount of dependencies one has to build. If someone doescargo add rerunand sees 200 dependencies being pulled in, they may very well abort.So: to use the SDK to only log to a file should be possible with a minimal set of dependencies; mostly just Arrow.
In the future we could add support for the native viewer streaming and .rrd file that is being written to by the SDK, thus still supporting
spawneven without WebSockets:.rrdfiles whilst they're still being written to #4056cargo tree -p rerun -e normalis good tool for working on this, as iscargo build --timings -p rerun --no-default-features