bound flume channel

Multi threade rendering results are now sent to a bounded channel. Meaning rendering threads will be temporarily blocked if they outpace the main thread transferring results to the final image. This wastes less memory. It is also potentially faster, since the main thread should be assigned priority.

Author: sloganking

src/render/function.rs

@@ -221,9 +221,14 @@ pub fn render(
         let mut line_num: u32 = 0;
         let mut longest_line_chars = 0;
         let mut background = None;
+        // An atomic integer used to tell threads which file to render next.
+        // Threads read a value and then incrment it.
+        // This is cheaper than creating a channel and sending the content to
+        // render via channel to each thread.
         let file_index = AtomicUsize::default();
+
         std::thread::scope(|scope| -> anyhow::Result<()> {
-            let (ttx, trx) = flume::unbounded();
+            let (ttx, trx) = flume::bounded(threads);
             for tid in 0..threads {
                 scope.spawn({
                     let ttx = ttx.clone();