This repo tries to give an overview over the landscape of the the different dbus implementations that exists in the rust ecosystem. The parameter I care about is executable size when using the library.
- https://github.com/KillingSpark/rustbus
- https://github.com/diwic/dbus-rs/ (bindings to C library)
- https://github.com/diwic/dbus-rs/tree/master/dbus-native
- https://gitlab.freedesktop.org/zeenix/zbus/
- https://github.com/Arnavion/dbus-pure
- https://github.com/srwalter/dbus-bytestream
- https://github.com/LinkTed/dbus-message-parser
- https://github.com/cmaves/async-rustbus
Adapted from timing benchmarks: https://github.com/KillingSpark/rust-dbus-comparisons
The dbus-message-parser does not provide any means of sending messages, so the result for it is pretty interesting.
I used rust 1.64.0 to run these benchmarks.
| Library | release with send | debug | release, marshal only |
|---|---|---|---|
| dbus-native | 356K | 6,4M | 312K |
| rustbus-async | 756K | 35M | 348K |
| zvariant | 728K | 21M | 428K |
| dbus-message-parser | 484K | 8,9M | 352K |
| dbus-bytestream | 352K | 6,5M | 336K |
| dbus-pure | 312K | 18M | 344K |
| rustbus | 484K | 11M | 332K |
| dbus-rs | 416K | 9,3M | 308K |
Why the hell did dbus-mesage-parser's size shoot up so much when adding sending?
dbus-native didn't, as expected, since everything is done inside C library.
dbus-pure shrank?
This benchmark was created to justify my prejudice against zbus and rust async ecosystem as a whole, and that was a success, their results are undeniably bad. Other than that, it's uncertain.