I shaved off 10 MiB from my binary in 2 hours!
I made a program using Macroquad, then I built it in release mode, the binary was 63 MiB in size.
So I used cargo vendor to have a better look at Macroquad and one of its dependencies, glam.
I then started to delete code, like, lots and lots of code(about 30_000 lines of code); none of it affected my main project, some of it became ‘dead_code’ just by removing the pub keyword.
The result is that my project was unaffected and the binary went down to 52 MiB.
Is there a way to automate removal of unneeded elements from dependencies? This is potentially huge.
Actually, dead code eliminination should do the trick, if you’re compiling a binary at least (same for a library I think, but there could be re-exports there). Did you compile in release mode?
To expand: Just configure whatever profile you’re using (
dev,release, …) to have link time optimization (lto) enabled:[profile.release] lto = "fat"This really doesn’t seem to do the trick, the binary’s still at 63MiB.
Also
"fat"andtrueare identical.Edit: I’m not sure I replied to the right person, ignore this.
The compiler doesn’t consider it to be dead code since it’s marked
pub.Sure, but isn’t this in a dependency? Can’t be reached when only importing your crate anyways? And if you’re building a binary, I don’t think this could really considered exported, is what I mean :)
Yes that’s exactly what I want. The compiler should stop considering it accessible.
I don’t recall what the default behavior is with the linker, but it might also benefit from at least thin LTO.
@Doods are you using
release? If you want the best dead-code elimination, you can also enable Link time optimisations (LTO): https://doc.rust-lang.org/cargo/reference/profiles.html#ltoYes, I am using both LTO and release mode, I can show you:
[profile.release] opt-level = 3 codegen-units = 1 panic = "abort" strip = true lto = truecargo build --release@Doods I’m surprised you can gain that much with that already enabled!
Should I bring it up to the ‘min-sized-rust’ working group or the forums or something?
Yes. This behavior seems strange, so either an explanation or investigation by a compiler dev seems like it would be helpful.
