build: apply optimizations in release mode

TODO: Potentially highlight more performance hotspots with
`#[optimize(speed)]`.

Cargo.toml

@@ -7,8 +7,23 @@ repository = "https://git.devcomp.xyz/DevComp/ssh-portfolio"
 authors = ["Erica Marigold <hi@devcomp.xyz>"]
 build = "build.rs"
 
+# Taken from https://github.com/microsoft/edit/blob/a3a6f5f/Cargo.toml#L20-L31
+# We use `opt-level = "s"` as it significantly reduces binary size.
+# We could then use the `#[optimize(speed)]` attribute for spot optimizations.
+# Unfortunately, that attribute currently doesn't work on intrinsics such as memset.
+[profile.release]
+codegen-units = 1           # Reduces binary size by ~2%
+debug = "full"              # No one needs an undebuggable release binary
+lto = true                  # Reduces binary size by ~14%
+opt-level = "s"             # Reduces binary size by ~25%
+split-debuginfo = "packed"  # Generates a separate *.dwp/*.dSYM so the binary can get stripped
+strip = "symbols"           # See split-debuginfo - allows us to drop the size by ~65%
+
+[profile.dev]
+opt-level = 2               # Optimize more than default, see https://github.com/RustCrypto/RSA/issues/144
+incremental = true          # Improves re-compile times
+
 [features]
-# TODO: CLI feature
 default = ["blog"]
 blog = [
     # Main deps
@@ -28,9 +43,6 @@ blog = [
     "dep:patch-crate"
 ]
 
-[profile.dev]
-opt-level = 2
-
 [package.metadata.patch]
 crates = ["ratatui-image"]
 

src/app.rs

@@ -107,6 +107,7 @@ impl App {
         })
     }
 
+    #[optimize(speed)]
     pub async fn run(
         &mut self,
         term: Arc<Mutex<Terminal>>,

src/config.rs

@@ -131,6 +131,7 @@ fn project_directory() -> Option<ProjectDirs> {
     ProjectDirs::from("xyz", "devcomp", env!("CARGO_PKG_NAME"))
 }
 
+#[optimize(speed)]
 fn private_key_deserialize<'de, D>(deserializer: D) -> Result<Vec<PrivateKey>, D::Error>
 where
     D: Deserializer<'de>, {

src/main.rs

@@ -1,3 +1,5 @@
+#![feature(optimize_attribute)]
+
 use std::net::SocketAddr;
 
 use clap::Parser as _;

src/ssh.rs

@@ -32,6 +32,7 @@ impl TermWriter {
         Self { session, channel, inner: CryptoVec::new() }
     }
 
+    #[optimize(speed)]
     fn flush_inner(&mut self) -> std::io::Result<()> {
         let handle = TokioHandle::current();
         handle.block_on(async move {
@@ -50,6 +51,7 @@ impl TermWriter {
 
 impl Write for TermWriter {
     #[instrument(skip(self, buf), level = "debug")]
+    #[optimize(speed)]
     fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
         tracing::trace!("Writing {} bytes into SSH terminal writer buffer", buf.len());
         self.inner.extend(buf);
@@ -57,6 +59,7 @@ impl Write for TermWriter {
     }
 
     #[instrument(skip(self), level = "trace")]
+    #[optimize(speed)]
     fn flush(&mut self) -> std::io::Result<()> {
         tracing::trace!("Flushing SSH terminal writer buffer");
         tokio::task::block_in_place(|| self.flush_inner())

src/tui/backend.rs

@@ -35,6 +35,7 @@ impl Backend for SshBackend {
         Ok(Size { width: self.dims.0, height: self.dims.1 })
     }
 
+    #[optimize(speed)]
     fn draw<'a, I>(&mut self, content: I) -> io::Result<()>
     where
         I: Iterator<Item = (u16, u16, &'a ratatui::buffer::Cell)>, {
@@ -71,6 +72,7 @@ impl Backend for SshBackend {
         })
     }
 
+    #[optimize(speed)]
     fn flush(&mut self) -> io::Result<()> {
         Backend::flush(&mut self.inner)
     }

src/tui/mod.rs

@@ -111,6 +111,7 @@ impl Tui {
         });
     }
 
+    #[optimize(speed)]
     async fn event_loop(
         status: Arc<RwLock<TuiStatus>>,
         event_tx: UnboundedSender<Event>,