perf: prevent repeated status fetches in large repos (#2824)

* perf: prevent repeated status fetches in large repos

Replace time-based cache invalidation with a generation counter.

The old `StatusParams` included a millisecond timestamp (tick) in its
hash, causing the cache to invalidate on every UI tick. For large repos
with millions of files, this led to repeated index loading and 5+ minute load times.

This change uses a different strategy to manage cache invalidation:
- Remove tick from StatusParams hash
- Add generation counter that increments after each fetch completes
- Include generation in cache hash

This ensures:
- No repeated fetches while one is already pending, making gitui usable
  in large repos
- New fetch starts immediately after completion, keeping gitui responsive
  in small repos
- External file changes will still be detected on the next fetch cycle

* fix changelog

---------

Co-authored-by: Daniel Stoll <dstoll@radix.trade>
Co-authored-by: extrawurst <776816+extrawurst@users.noreply.github.com>
Co-authored-by: extrawurst <mail@rusticorn.com>

Author: 3 people

CHANGELOG.md

@@ -8,14 +8,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## Unreleased
 
 ### Changed
-<<<<<<< feat/version-message
 * improve `gitui --version` message [[@hlsxx](https://github.com/hlsxx)] ([#2838](https://github.com/gitui-org/gitui/issues/2838))
-=======
 * rust msrv bumped to `1.88`
 
 ### Fixed
+* fix extremely slow status loading in large repositories by replacing time-based cache invalidation with generation counter [[@DannyStoll1](https://github.com/DannyStoll1)] ([#2823](https://github.com/gitui-org/gitui/issues/2823))
 * fix panic when renaming or updating remote URL with no remotes configured [[@xvchris](https://github.com/xvchris)] ([#2868](https://github.com/gitui-org/gitui/issues/2868))
->>>>>>> master
 
 ## [0.28.0] - 2025-12-14
 

asyncgit/src/status.rs

@@ -10,19 +10,11 @@ use crossbeam_channel::Sender;
 use std::{
 	hash::Hash,
 	sync::{
-		atomic::{AtomicUsize, Ordering},
+		atomic::{AtomicU64, AtomicUsize, Ordering},
 		Arc, Mutex,
 	},
-	time::{SystemTime, UNIX_EPOCH},
 };
 
-fn current_tick() -> u128 {
-	SystemTime::now()
-		.duration_since(UNIX_EPOCH)
-		.expect("time before unix epoch!")
-		.as_millis()
-}
-
 #[derive(Default, Hash, Clone)]
 pub struct Status {
 	pub items: Vec<StatusItem>,
@@ -31,19 +23,17 @@ pub struct Status {
 ///
 #[derive(Default, Hash, Copy, Clone, PartialEq, Eq)]
 pub struct StatusParams {
-	tick: u128,
 	status_type: StatusType,
 	config: Option<ShowUntrackedFilesConfig>,
 }
 
 impl StatusParams {
 	///
-	pub fn new(
+	pub const fn new(
 		status_type: StatusType,
 		config: Option<ShowUntrackedFilesConfig>,
 	) -> Self {
 		Self {
-			tick: current_tick(),
 			status_type,
 			config,
 		}
@@ -59,6 +49,8 @@ pub struct AsyncStatus {
 	sender: Sender<AsyncGitNotification>,
 	pending: Arc<AtomicUsize>,
 	repo: RepoPath,
+	/// Counter that increments after each completed fetch.
+	generation: Arc<AtomicU64>,
 }
 
 impl AsyncStatus {
@@ -73,6 +65,7 @@ impl AsyncStatus {
 			last: Arc::new(Mutex::new(Status::default())),
 			sender,
 			pending: Arc::new(AtomicUsize::new(0)),
+			generation: Arc::new(AtomicU64::new(0)),
 		}
 	}
 
@@ -97,12 +90,14 @@ impl AsyncStatus {
 			return Ok(None);
 		}
 
-		let hash_request = hash(&params);
+		let generation = self.generation.load(Ordering::Relaxed);
+		let hash_request = hash(&(params, generation));
 
 		log::trace!(
-			"request: [hash: {}] (type: {:?})",
+			"request: [hash: {}] (type: {:?}, gen: {})",
 			hash_request,
 			params.status_type,
+			generation,
 		);
 
 		{
@@ -118,6 +113,7 @@ impl AsyncStatus {
 
 		let arc_current = Arc::clone(&self.current);
 		let arc_last = Arc::clone(&self.last);
+		let arc_generation = Arc::clone(&self.generation);
 		let sender = self.sender.clone();
 		let arc_pending = Arc::clone(&self.pending);
 		let status_type = params.status_type;
@@ -138,6 +134,8 @@ impl AsyncStatus {
 				log::error!("fetch_helper: {e}");
 			}
 
+			// Increment generation to invalidate cache for next request
+			arc_generation.fetch_add(1, Ordering::Relaxed);
 			arc_pending.fetch_sub(1, Ordering::Relaxed);
 
 			sender