perf(vcs): filter the Git log with git log instead of a background search index (#1537)

* perf(vcs): filter the Git log with git log instead of a background search index

The log built a persistent inverted index of every commit's message, authors and
changed paths in the background, with a progress bar above the commit list, and
walked trees itself to filter by path. Building it took minutes on large
repositories, and filters were slow until it finished.

- Text: plain text uses `git log --fixed-strings --grep` (with
  `--regexp-ignore-case` unless Match Case is on); text of 4+ hex digits also
  matches hash prefixes. Regex filters still match in-process.
- Paths: `git log -- :(literal)<paths>` replaces the in-process path walk and
  its per-commit tree cache. It uses git's history simplification and, when the
  repository has a changed-path commit-graph, its Bloom filters. Glob characters
  in a path are literal.
- Authors match against the commit metadata cache.
- Removed the search tables, `index_log`, the `vcs_log_index` command and the
  indexing progress bar. The commit metadata cache stays.

* docs(contributing): ask for screenshots only when a change alters what users see

Removing a feature together with its UI, copy-only and non-visual changes now
describe what was checked instead of attaching before/after screenshots.

* fix(vcs): match case-insensitive non-ASCII Git log text in process

git folds case only for ASCII under the C locale it runs with, so a
case-insensitive filter like "ärger" missed "Ärger". Such text is now
matched against commit messages directly. Path and text selection also
ignore log.showSignature and log.follow from the user's config.

.github/pull_request_template.md

@@ -9,7 +9,8 @@
 
 <!-- Required. Commands you actually ran and their results; what else this could
      affect and how you checked it still works.
-     UI changes: before/after screenshots. Multi-step interactions: a recording.
+     Visible UI changes: before/after screenshots (not for removed UI or copy-only
+     changes). Multi-step interactions: a recording.
      Backend changes: logs or command output. -->
 
 ## Review focus

AGENTS.md

@@ -51,7 +51,8 @@ Visual style: dark, low-contrast greys, one cool accent
 
 Follow `CONTRIBUTING.md` and `.github/pull_request_template.md`: one problem per PR
 (`Closes #<n>` when it resolves an issue), reproduce first, keep the diff focused, short plain description, and a
-**How it was verified** section with commands run, results and (for UI) before/after screenshots.
+**How it was verified** section with commands run, results and, when the change alters what users
+see, before/after screenshots (not for UI removed along with a feature, copy-only or non-visual changes).
 Unclear product decisions → `needs-decision` and stop.
 
 ## Skills

CONTRIBUTING.md

@@ -25,8 +25,9 @@ Project layout, coding rules, commands and checks are in [AGENTS.md](AGENTS.md).
 - Fill in the template. Keep the description short and in plain words. If it
   doesn't fit in a few sentences, the PR is probably too big.
 - **How it was verified** is required: the commands you ran and their results,
-  and what else the change could affect. UI changes need before/after screenshots;
-  multi-step interactions need a recording.
+  and what else the change could affect. Changes to what users see need before/after
+  screenshots; multi-step interactions need a recording. UI that goes away with a removed
+  feature, copy-only and non-visual changes don't; say what you checked instead.
 - If a change needs a product decision you can't make, ask in the issue and stop
   rather than guessing.
 

crates/ide-host/src/ops/vcs/commands.rs

@@ -405,28 +405,6 @@ pub async fn vcs_log_details_batch(
     result
 }
 
-pub async fn vcs_log_index(
-    host: &Host,
-    window: &str,
-    repo: Option<PathBuf>,
-    request: Option<RequestId>,
-) -> IdeResult<LogIndexProgress> {
-    let project = LogAccess::new(host, window, repo.as_deref());
-    let cancel =
-        request.map_or_else(ide_core::CancellationToken::new, |id| host.begin_project_read_request(window, id));
-    let result = blocking(move || {
-        project.with_log_repo_at(repo.as_deref(), |r| {
-            r.enable_log_index(&project.cache);
-            if request.is_some() { r.index_log(&cancel) } else { Ok(r.log_index_progress()) }
-        })
-    })
-    .await;
-    if let Some(request) = request {
-        host.end_request(request);
-    }
-    result
-}
-
 pub async fn vcs_revision_children(
     host: &Host,
     window: &str,
@@ -1222,10 +1200,6 @@ pub(crate) fn register(d: &mut Dispatcher) {
     d.add("vcs_log_details_batch", |c| async move {
         vcs_log_details_batch(&c.host, &c.window, c.arg("repo")?, c.arg("revisions")?, c.arg("request")?).await
     });
-    d.add(
-        "vcs_log_index",
-        |c| async move { vcs_log_index(&c.host, &c.window, c.arg("repo")?, c.arg("request")?).await },
-    );
     d.add("vcs_revision_children", |c| async move {
         vcs_revision_children(&c.host, &c.window, c.arg("repo")?, c.arg("revision")?, c.arg("directory")?).await
     });

crates/ide-host/src/trust.rs

@@ -316,7 +316,6 @@ pub fn restricted_command_allowed(cmd: &str) -> bool {
             | "vcs_log_users"
             | "vcs_log_details"
             | "vcs_log_details_batch"
-            | "vcs_log_index"
             | "vcs_revision_children"
             | "vcs_revision_files"
             | "vcs_open_external_log"

crates/ide-vcs/src/history.rs

@@ -9,44 +9,88 @@ use std::collections::{HashMap, HashSet};
 
 /// Log "Text or hash" filter: the message (substring or regex), or a hash prefix.
 enum TextFilter {
-    Plain { needle: String, match_case: bool },
+    /// Commits `git log --grep` selected, or whose id starts with `hash`.
+    Selected {
+        ids: HashSet<ObjectId>,
+        hash: Option<String>,
+    },
+    /// Case-insensitive non-ASCII text: git folds case only for ASCII under the C locale.
+    Folded(String),
     Regex(regex_lite::Regex),
 }
 
 impl TextFilter {
-    fn new(text: &str, regex: bool, match_case: bool) -> IdeResult<Option<Self>> {
+    fn new(
+        r: &Repo,
+        scopes: &[String],
+        q: &LogQuery,
+        text: &str,
+        cancel: &CancellationToken,
+    ) -> IdeResult<Option<Self>> {
         let text = text.trim();
         if text.is_empty() {
             return Ok(None);
         }
-        if regex {
+        if q.text_regex {
             let re = regex_lite::RegexBuilder::new(text)
-                .case_insensitive(!match_case)
+                .case_insensitive(!q.text_match_case)
                 .multi_line(true)
                 .build()
                 .map_err(|e| IdeError::invalid(format!("invalid regular expression: {e}")))?;
             return Ok(Some(Self::Regex(re)));
         }
-        let needle = if match_case { text.to_string() } else { text.to_lowercase() };
-        Ok(Some(Self::Plain { needle, match_case }))
+        if !q.text_match_case && !text.is_ascii() {
+            return Ok(Some(Self::Folded(text.to_lowercase())));
+        }
+        let mut filter = vec!["--fixed-strings".to_string(), format!("--grep={text}")];
+        if !q.text_match_case {
+            filter.push("--regexp-ignore-case".into());
+        }
+        let ids = git_select(r, scopes, &filter, &[], q.first_parent, cancel)?;
+        let hash = (text.len() >= 4 && text.bytes().all(|b| b.is_ascii_hexdigit())).then(|| text.to_ascii_lowercase());
+        Ok(Some(Self::Selected { ids, hash }))
     }
 
-    fn matches(&self, message: &str, id: &ObjectId) -> bool {
-        match self {
-            Self::Regex(re) => re.is_match(message),
-            Self::Plain { needle, match_case } => {
-                let hash = needle.len() >= 4 && needle.bytes().all(|b| b.is_ascii_hexdigit());
-                if hash && id.to_string().starts_with(&needle.to_ascii_lowercase()) {
-                    return true;
-                }
-                if *match_case {
-                    message.contains(needle.as_str())
-                } else {
-                    message.to_lowercase().contains(needle.as_str())
-                }
+    fn matches(&self, repo: &gix::Repository, index: &mut crate::log_index::LogIndex, id: ObjectId) -> IdeResult<bool> {
+        Ok(match self {
+            Self::Regex(re) => re.is_match(&index.message(repo, id)?),
+            Self::Folded(needle) => index.message(repo, id)?.to_lowercase().contains(needle.as_str()),
+            Self::Selected { ids, hash } => {
+                ids.contains(&id) || hash.as_ref().is_some_and(|h| id.to_string().starts_with(h.as_str()))
             }
+        })
+    }
+}
+
+/// Commits git selects from each scope with `filter`. Path limiting uses git's history
+/// simplification and, when the repository has them, changed-path Bloom filters.
+fn git_select(
+    r: &Repo,
+    scopes: &[String],
+    filter: &[String],
+    paths: &[String],
+    first_parent: bool,
+    cancel: &CancellationToken,
+) -> IdeResult<HashSet<ObjectId>> {
+    let mut ids = HashSet::new();
+    for scope in scopes {
+        let mut args =
+            vec!["log".to_string(), "--format=%H".into(), "--no-show-signature".into(), "--no-follow".into()];
+        if first_parent {
+            args.push("--first-parent".into());
+        }
+        args.extend(filter.iter().cloned());
+        args.push("--end-of-options".into());
+        args.extend(scope.split_whitespace().map(str::to_string));
+        args.push("--".into());
+        args.extend(paths.iter().map(|p| format!(":(literal){p}")));
+        let output =
+            crate::cli::run_cancellable(r.root(), &args.iter().map(String::as_str).collect::<Vec<_>>(), cancel)?;
+        for line in output.stdout_str().lines() {
+            ids.insert(ObjectId::from_hex(line.as_bytes()).map_err(gx)?);
         }
     }
+    Ok(ids)
 }
 
 pub(crate) fn log(r: &Repo, q: &LogQuery, cancel: &CancellationToken) -> IdeResult<Vec<Commit>> {
@@ -173,87 +217,71 @@ fn log_indexed(
     } else {
         None
     };
+    let scopes = if let Some(range) = range {
+        vec![range.to_string()]
+    } else {
+        let mut scopes = ranges.iter().map(|s| s.to_string()).collect::<Vec<_>>();
+        if !tips.is_empty() {
+            scopes.push(tips.iter().map(ToString::to_string).collect::<Vec<_>>().join(" "));
+        }
+        scopes
+    };
     let mut history_paths = HashMap::new();
     let path_ids = if paths.is_empty() {
         None
-    } else {
+    } else if q.follow && paths.len() == 1 {
         let mut ids = HashSet::new();
-        if q.follow && paths.len() == 1 {
-            let rel = &paths[0];
-            let mut args = vec![
-                "log".to_string(),
-                "--format=%x00%H".into(),
-                "--name-status".into(),
-                "-z".into(),
-                "--follow".into(),
-                "--topo-order".into(),
-            ];
-            if q.first_parent {
-                args.push("--first-parent".into());
-            }
-            args.push("--end-of-options".into());
-            let scopes = if let Some(range) = range {
-                vec![range.to_string()]
-            } else {
-                let mut scopes = ranges.iter().map(|s| s.to_string()).collect::<Vec<_>>();
-                if !tips.is_empty() {
-                    scopes.push(tips.iter().map(ToString::to_string).collect::<Vec<_>>().join(" "));
-                }
-                scopes
-            };
-            for scope in scopes {
-                let mut args = args.clone();
-                args.extend(scope.split_whitespace().map(str::to_string));
-                args.extend(["--".into(), rel.clone()]);
-                let output = crate::cli::run_cancellable(
-                    r.root(),
-                    &args.iter().map(String::as_str).collect::<Vec<_>>(),
-                    cancel,
-                )?;
-                let text = output.stdout_str();
-                let mut tokens = text.split('\0').peekable();
-                let mut current_path = rel.clone();
-                while let Some(token) = tokens.next() {
-                    let token = token.trim_start_matches('\n');
-                    if let Ok(id) = ObjectId::from_hex(token.as_bytes()) {
-                        ids.insert(id);
-                        history_paths.insert(id, current_path.clone());
-                    } else if token.starts_with('R') && token[1..].bytes().all(|b| b.is_ascii_digit()) {
-                        if let (Some(old), Some(new)) = (tokens.next(), tokens.next())
-                            && new == current_path
-                        {
-                            current_path = old.to_string();
-                        }
-                    } else if ["A", "D", "M", "T", "U"].contains(&token) {
-                        tokens.next();
+        let rel = &paths[0];
+        let mut args = vec![
+            "log".to_string(),
+            "--format=%x00%H".into(),
+            "--name-status".into(),
+            "-z".into(),
+            "--follow".into(),
+            "--topo-order".into(),
+        ];
+        if q.first_parent {
+            args.push("--first-parent".into());
+        }
+        args.push("--end-of-options".into());
+        for scope in &scopes {
+            let mut args = args.clone();
+            args.extend(scope.split_whitespace().map(str::to_string));
+            args.extend(["--".into(), rel.clone()]);
+            let output =
+                crate::cli::run_cancellable(r.root(), &args.iter().map(String::as_str).collect::<Vec<_>>(), cancel)?;
+            let text = output.stdout_str();
+            let mut tokens = text.split('\0').peekable();
+            let mut current_path = rel.clone();
+            while let Some(token) = tokens.next() {
+                let token = token.trim_start_matches('\n');
+                if let Ok(id) = ObjectId::from_hex(token.as_bytes()) {
+                    ids.insert(id);
+                    history_paths.insert(id, current_path.clone());
+                } else if token.starts_with('R') && token[1..].bytes().all(|b| b.is_ascii_digit()) {
+                    if let (Some(old), Some(new)) = (tokens.next(), tokens.next())
+                        && new == current_path
+                    {
+                        current_path = old.to_string();
                     }
+                } else if ["A", "D", "M", "T", "U"].contains(&token) {
+                    tokens.next();
                 }
             }
-        } else {
-            let starts = range_ids.as_ref().cloned().unwrap_or_else(|| tips.clone());
-            ids.extend(
-                PathWalk::new(&repo, starts, paths.clone(), q.first_parent, cancel, index)?
-                    .map(|item| item.map(|(id, _)| id))
-                    .collect::<IdeResult<HashSet<_>>>()?,
-            );
         }
         Some(ids)
+    } else {
+        Some(git_select(r, &scopes, &[], &paths, q.first_parent, cancel)?)
     };
     let authors: Vec<_> =
         q.author.iter().chain(&q.authors).map(|a| a.trim().to_lowercase()).filter(|a| !a.is_empty()).collect();
-    let user_candidates = if q.authors_exact && !authors.is_empty() { index.user_candidates(&authors) } else { None };
-    let path_candidates = if !q.follow && !paths.is_empty() { index.path_candidates(&paths) } else { None };
-    let text_candidates = q.text.as_deref().filter(|_| !q.text_regex).and_then(|s| index.message_candidates(s));
     let mut visibility = Visibility {
         speed_refs: &decorations,
-        text_candidates,
-        user_candidates,
-        path_candidates,
         query: q,
         path_ids,
         range_ids: range_ids.as_ref().map(|ids| ids.iter().copied().collect()),
         authors,
-        text: q.text.as_deref().map(|t| TextFilter::new(t, q.text_regex, q.text_match_case)).transpose()?.flatten(),
+        text: q.text.as_deref().map(|t| TextFilter::new(r, &scopes, q, t, cancel)).transpose()?.flatten(),
         matches: HashMap::new(),
         projected: HashMap::new(),
         cancel,
@@ -584,9 +612,6 @@ struct Visibility<'a> {
     range_ids: Option<HashSet<ObjectId>>,
     authors: Vec<String>,
     text: Option<TextFilter>,
-    text_candidates: Option<HashSet<String>>,
-    user_candidates: Option<HashSet<String>>,
-    path_candidates: Option<HashSet<String>>,
     matches: HashMap<ObjectId, bool>,
     projected: HashMap<ObjectId, Vec<ObjectId>>,
     cancel: &'a CancellationToken,
@@ -602,10 +627,7 @@ impl Visibility<'_> {
         if let Some(shown) = self.matches.get(&id) {
             return Ok(*shown);
         }
-        let indexed = self.index.indexed(&id.to_string());
-        let shown = if (indexed && self.user_candidates.as_ref().is_some_and(|ids| !ids.contains(&id.to_string())))
-            || (indexed && self.path_candidates.as_ref().is_some_and(|ids| !ids.contains(&id.to_string())))
-            || !repo.has_object(id)
+        let shown = if !repo.has_object(id)
             || self.path_ids.as_ref().is_some_and(|ids| !ids.contains(&id))
             || self.range_ids.as_ref().is_some_and(|ids| !ids.contains(&id))
         {
@@ -648,15 +670,7 @@ impl Visibility<'_> {
                         }
                     }))
                 && match &self.text {
-                    Some(text) => {
-                        self.text_candidates.as_ref().is_none_or(|ids| {
-                            ids.contains(&id.to_string())
-                                || !self.index.indexed(&id.to_string())
-                                || id.to_string().starts_with(
-                                    self.query.text.as_deref().unwrap_or("").trim().to_ascii_lowercase().as_str(),
-                                )
-                        }) && text.matches(&self.index.message(repo, id)?, &id)
-                    }
+                    Some(text) => text.matches(repo, self.index, id)?,
                     None => true,
                 }
         };
@@ -865,113 +879,6 @@ pub(crate) fn decorations(repo: &gix::Repository) -> IdeResult<HashMap<ObjectId,
     Ok(map)
 }
 
-/// Commits touching a path, newest first, with git's default history simplification:
-/// a commit is shown when its entry for the path differs from every parent, and at a
-/// merge that kept one parent's version only that parent's line of history is followed.
-struct PathWalk<'r> {
-    repo: &'r gix::Repository,
-    paths: Vec<String>,
-    queue: std::collections::BinaryHeap<(i64, u64, ObjectId)>,
-    seen: HashSet<ObjectId>,
-    index: &'r mut crate::log_index::LogIndex,
-    seq: u64,
-    first_parent: bool,
-    cancel: &'r CancellationToken,
-    shallow: HashSet<ObjectId>,
-}
-
-impl<'r> PathWalk<'r> {
-    fn new(
-        repo: &'r gix::Repository,
-        tips: Vec<ObjectId>,
-        paths: Vec<String>,
-        first_parent: bool,
-        cancel: &'r CancellationToken,
-        index: &'r mut crate::log_index::LogIndex,
-    ) -> IdeResult<Self> {
-        let mut walk = PathWalk {
-            repo,
-            paths,
-            queue: Default::default(),
-            seen: Default::default(),
-            index,
-            seq: 0,
-            first_parent,
-            cancel,
-            shallow: repo.shallow_commits().map_err(gx)?.map(|ids| ids.iter().copied().collect()).unwrap_or_default(),
-        };
-        for tip in tips {
-            walk.push(tip)?;
-        }
-        Ok(walk)
-    }
-
-    fn push(&mut self, id: ObjectId) -> IdeResult<()> {
-        if self.cancel.is_cancelled() {
-            return Err(IdeError::cancelled());
-        }
-        if !self.seen.insert(id) || !self.repo.has_object(id) {
-            return Ok(());
-        }
-        let time = self.repo.find_commit(id).map_err(gx)?.time().map(|t| t.seconds).unwrap_or(0);
-        // Max-heap on time; among equal times, earlier insertion first.
-        self.seq += 1;
-        self.queue.push((time, u64::MAX - self.seq, id));
-        Ok(())
-    }
-
-    fn step(&mut self) -> IdeResult<Option<(ObjectId, Vec<ObjectId>)>> {
-        while let Some((_, _, id)) = self.queue.pop() {
-            let commit = self.repo.find_commit(id).map_err(gx)?;
-            let parents: Vec<ObjectId> = commit.parent_ids().map(|p| p.detach()).collect();
-            // Parents cut off by a shallow boundary are treated as absent.
-            let present: Vec<ObjectId> = parents
-                .iter()
-                .copied()
-                .take(if self.shallow.contains(&id) { 0 } else { usize::MAX })
-                .filter(|p| self.repo.has_object(p))
-                .take(if self.first_parent { 1 } else { usize::MAX })
-                .collect();
-            let mine = self
-                .paths
-                .iter()
-                .map(|path| self.index.path_entry(self.repo, id, path))
-                .collect::<IdeResult<Vec<_>>>()?;
-            let mut same_parent = None;
-            for p in &present {
-                let theirs = self
-                    .paths
-                    .iter()
-                    .map(|path| self.index.path_entry(self.repo, *p, path))
-                    .collect::<IdeResult<Vec<_>>>()?;
-                if theirs == mine {
-                    same_parent = Some(*p);
-                    break;
-                }
-            }
-            match same_parent {
-                Some(p) => self.push(p)?,
-                None => {
-                    for p in &present {
-                        self.push(*p)?;
-                    }
-                    if !present.is_empty() || mine.iter().any(Option::is_some) {
-                        return Ok(Some((id, parents)));
-                    }
-                }
-            }
-        }
-        Ok(None)
-    }
-}
-
-impl Iterator for PathWalk<'_> {
-    type Item = IdeResult<(ObjectId, Vec<ObjectId>)>;
-    fn next(&mut self) -> Option<Self::Item> {
-        self.step().transpose()
-    }
-}
-
 pub(crate) fn commit_files(r: &Repo, rev: &str) -> IdeResult<Vec<ChangedFile>> {
     let repo = r.gix()?;
     Ok(first_parent_changes(&repo, rev)?.files)

crates/ide-vcs/src/lib.rs

@@ -162,14 +162,6 @@ pub struct LogDetails {
     pub fingerprint: String,
 }
 
-#[derive(Clone, Debug, Default, Serialize, Deserialize, Type)]
-#[serde(rename_all = "camelCase")]
-pub struct LogIndexProgress {
-    pub indexed: u32,
-    pub total: u32,
-    pub running: bool,
-}
-
 /// Order of the Git Log (Graph Options › Sort).
 #[derive(Clone, Copy, Debug, Default, PartialEq, Eq, Serialize, Deserialize, Type)]
 #[serde(rename_all = "camelCase")]
@@ -672,7 +664,6 @@ pub struct Repo {
     inner: std::sync::Mutex<repo::Cache>,
     ignored: std::sync::RwLock<IgnoredNames>,
     log_index: std::sync::Mutex<log_index::LogIndex>,
-    index_progress: std::sync::Mutex<LogIndexProgress>,
     log_signatures: std::sync::Mutex<std::collections::HashMap<String, LogDetails>>,
 }
 
@@ -844,59 +835,6 @@ impl Repo {
             .into_iter()
             .collect())
     }
-    pub fn log_index_progress(&self) -> LogIndexProgress {
-        self.index_progress.lock().unwrap_or_else(|p| p.into_inner()).clone()
-    }
-
-    /// Incremental persistent inverted index; completed objects survive cancellation/restart.
-    pub fn index_log(&self, cancel: &CancellationToken) -> IdeResult<LogIndexProgress> {
-        {
-            let mut progress = self.index_progress.lock().unwrap_or_else(|p| p.into_inner());
-            if progress.running {
-                return Ok(progress.clone());
-            }
-            progress.running = true;
-        }
-        let result = (|| {
-            let repo = self.gix()?;
-            let out = cli::run_cancellable(self.root(), &["rev-list", "--all", "--topo-order"], cancel)?;
-            let ids: Vec<_> = out
-                .stdout_str()
-                .lines()
-                .map(|id| gix::ObjectId::from_hex(id.as_bytes()).map_err(repo::gx))
-                .collect::<IdeResult<_>>()?;
-            {
-                let mut p = self.index_progress.lock().unwrap_or_else(|p| p.into_inner());
-                p.total = ids.len() as u32;
-                p.indexed = 0;
-            }
-            for (i, id) in ids.iter().enumerate() {
-                if cancel.is_cancelled() {
-                    return Err(ide_core::IdeError::cancelled());
-                }
-                let indexed = self.log_index.lock().unwrap_or_else(|p| p.into_inner()).indexed(&id.to_string());
-                if !indexed {
-                    let files = self.commit_files(&id.to_string())?;
-                    let paths =
-                        files.into_iter().flat_map(|f| std::iter::once(f.path).chain(f.old_path)).collect::<Vec<_>>();
-                    let mut index = self.log_index.lock().unwrap_or_else(|p| p.into_inner());
-                    let c = index.get(&repo, *id)?;
-                    let message = index.message(&repo, *id)?;
-                    index.index_commit(&c.id, &message, &c.author, &c.email, &paths);
-                    if i % 100 == 0 {
-                        index.flush();
-                    }
-                }
-                self.index_progress.lock().unwrap_or_else(|p| p.into_inner()).indexed = i as u32 + 1;
-            }
-            Ok(())
-        })();
-        self.log_index.lock().unwrap_or_else(|p| p.into_inner()).flush();
-        self.index_progress.lock().unwrap_or_else(|p| p.into_inner()).running = false;
-        result?;
-        Ok(self.log_index_progress())
-    }
-
     pub fn log_details(&self, revision: &str) -> IdeResult<LogDetails> {
         self.log_details_cancellable(revision, &CancellationToken::new())
     }

crates/ide-vcs/src/log_index.rs

@@ -6,7 +6,7 @@ use crate::repo::{gx, to_string};
 use gix::ObjectId;
 use gix::bstr::ByteSlice;
 use ide_core::IdeResult;
-use redb::{Database, MultimapTableDefinition, ReadableDatabase, TableDefinition};
+use redb::{Database, ReadableDatabase, TableDefinition};
 use std::collections::{HashMap, HashSet};
 use std::path::Path;
 use std::sync::{Arc, Mutex, OnceLock, Weak};
@@ -14,9 +14,6 @@ use std::time::SystemTime;
 
 const COMMITS: TableDefinition<&str, &[u8]> = TableDefinition::new("summaries-v2");
 const MESSAGES: TableDefinition<&str, &str> = TableDefinition::new("messages-v1");
-const PATHS: TableDefinition<&str, &[u8]> = TableDefinition::new("paths-v2");
-const POSTINGS: MultimapTableDefinition<&str, &str> = MultimapTableDefinition::new("search-v1");
-const INDEXED: TableDefinition<&str, bool> = TableDefinition::new("indexed-v1");
 
 pub(crate) struct IndexedView {
     pub commits: Vec<Commit>,
@@ -36,9 +33,6 @@ pub(crate) struct LogIndex {
     memory: HashMap<ObjectId, Arc<Commit>>,
     dirty: Vec<Arc<Commit>>,
     dirty_messages: Vec<(String, String)>,
-    dirty_search: Vec<(String, Vec<String>)>,
-    path_memory: HashMap<(ObjectId, String), Option<(ObjectId, u16)>>,
-    dirty_paths: Vec<(String, Option<(String, u16)>)>,
     pub head: Option<(ObjectId, Option<SystemTime>)>,
     pub reachable: HashSet<ObjectId>,
     pub views: HashMap<String, Arc<IndexedView>>,
@@ -92,91 +86,6 @@ impl LogIndex {
         }
         Ok(ids)
     }
-    pub fn indexed(&self, id: &str) -> bool {
-        self.db
-            .as_ref()
-            .and_then(|db| {
-                let txn = db.begin_read().ok()?;
-                let table = txn.open_table(INDEXED).ok()?;
-                Some(table.get(id).ok()??.value())
-            })
-            .unwrap_or(false)
-    }
-
-    pub fn index_commit(&mut self, id: &str, message: &str, author: &str, email: &str, paths: &[String]) {
-        let mut keys: HashSet<String> = Self::grams(message).into_iter().map(|s| format!("m:{s}")).collect();
-        for user in [author, email, email.split('@').next().unwrap_or(email)] {
-            keys.insert(format!("u:{}", normalize_user(user)));
-        }
-        for path in paths {
-            keys.insert(format!("p:{path}"));
-        }
-        self.dirty_search.push((id.into(), keys.into_iter().collect()));
-    }
-
-    fn grams(text: &str) -> Vec<String> {
-        let chars: Vec<_> = text.to_lowercase().chars().collect();
-        chars.windows(3).map(|s| s.iter().collect()).collect::<HashSet<_>>().into_iter().collect()
-    }
-
-    /// Missing/unindexed commits are always checked directly by the caller.
-    pub fn message_candidates(&self, text: &str) -> Option<HashSet<String>> {
-        let grams = Self::grams(text);
-        if grams.is_empty() {
-            return None;
-        }
-        let db = self.db.as_ref()?;
-        let txn = db.begin_read().ok()?;
-        let table = txn.open_multimap_table(POSTINGS).ok()?;
-        let mut candidates: Option<HashSet<String>> = None;
-        for gram in grams {
-            let values = table.get(format!("m:{gram}").as_str()).ok()?;
-            let ids: HashSet<_> = values.filter_map(Result::ok).map(|s| s.value().to_string()).collect();
-            candidates = Some(match candidates {
-                Some(old) => old.intersection(&ids).cloned().collect(),
-                None => ids,
-            });
-            if candidates.as_ref().is_some_and(HashSet::is_empty) {
-                break;
-            }
-        }
-        Some(candidates.unwrap_or_default())
-    }
-
-    pub fn user_candidates(&self, authors: &[String]) -> Option<HashSet<String>> {
-        let txn = self.db.as_ref()?.begin_read().ok()?;
-        let table = txn.open_multimap_table(POSTINGS).ok()?;
-        let mut ids = HashSet::new();
-        for author in authors {
-            let name = author.split('<').next().unwrap_or(author).trim();
-            let email = author.split('<').nth(1).map(|s| s.trim_end_matches('>')).unwrap_or(author);
-            for alias in [name, email, email.split('@').next().unwrap_or(email)] {
-                for value in table.get(format!("u:{}", normalize_user(alias)).as_str()).ok()?.filter_map(Result::ok) {
-                    ids.insert(value.value().to_string());
-                }
-            }
-        }
-        Some(ids)
-    }
-
-    pub fn path_candidates(&self, paths: &[String]) -> Option<HashSet<String>> {
-        let txn = self.db.as_ref()?.begin_read().ok()?;
-        let table = txn.open_multimap_table(POSTINGS).ok()?;
-        let mut ids = HashSet::new();
-        for path in paths {
-            for value in table.get(format!("p:{path}").as_str()).ok()?.filter_map(Result::ok) {
-                ids.insert(value.value().to_string());
-            }
-            let prefix = format!("p:{path}/");
-            let end = format!("{prefix}{}", char::MAX);
-            for row in table.range(prefix.as_str()..end.as_str()).ok()?.filter_map(Result::ok) {
-                for value in row.1.filter_map(Result::ok) {
-                    ids.insert(value.value().to_string());
-                }
-            }
-        }
-        Some(ids)
-    }
     pub fn open(&mut self, dir: &Path) {
         if self.opened {
             return;
@@ -272,11 +181,7 @@ impl LogIndex {
     }
 
     pub fn flush(&mut self) {
-        if self.dirty.is_empty()
-            && self.dirty_paths.is_empty()
-            && self.dirty_messages.is_empty()
-            && self.dirty_search.is_empty()
-        {
+        if self.dirty.is_empty() && self.dirty_messages.is_empty() {
             return;
         }
         let result = (|| -> Result<(), Box<dyn std::error::Error>> {
@@ -291,29 +196,12 @@ impl LogIndex {
                     table.insert(commit.id.as_str(), bytes.as_slice())?;
                 }
             }
-            {
-                let mut table = txn.open_table(PATHS)?;
-                for (key, value) in &self.dirty_paths {
-                    let bytes = serde_json::to_vec(value)?;
-                    table.insert(key.as_str(), bytes.as_slice())?;
-                }
-            }
             {
                 let mut table = txn.open_table(MESSAGES)?;
                 for (id, message) in &self.dirty_messages {
                     table.insert(id.as_str(), message.as_str())?;
                 }
             }
-            {
-                let mut postings = txn.open_multimap_table(POSTINGS)?;
-                let mut indexed = txn.open_table(INDEXED)?;
-                for (id, keys) in &self.dirty_search {
-                    for key in keys {
-                        postings.insert(key.as_str(), id.as_str())?;
-                    }
-                    indexed.insert(id.as_str(), true)?;
-                }
-            }
             txn.commit()?;
             Ok(())
         })();
@@ -321,52 +209,7 @@ impl LogIndex {
             tracing::warn!("Git log cache write failed: {error}");
         }
         self.dirty_messages.clear();
-        self.dirty_search.clear();
         self.dirty.clear();
-        self.dirty_paths.clear();
-    }
-
-    pub fn path_entry(
-        &mut self,
-        repo: &gix::Repository,
-        id: ObjectId,
-        path: &str,
-    ) -> IdeResult<Option<(ObjectId, u16)>> {
-        let memory_key = (id, path.to_string());
-        if let Some(value) = self.path_memory.get(&memory_key) {
-            return Ok(*value);
-        }
-        let key = format!("{id}\0{path}");
-        let cached = self.db.as_ref().and_then(|db| {
-            let txn = db.begin_read().ok()?;
-            let table = txn.open_table(PATHS).ok()?;
-            let value = table.get(key.as_str()).ok()??;
-            serde_json::from_slice::<Option<(String, u16)>>(value.value()).ok()
-        });
-        let cached = match cached {
-            Some(None) => Some(None),
-            Some(Some((hex, mode))) => ObjectId::from_hex(hex.as_bytes()).ok().map(|id| Some((id, mode))),
-            None => None,
-        };
-        let value = if let Some(value) = cached {
-            value
-        } else {
-            let tree = repo.find_commit(id).map_err(gx)?.tree().map_err(gx)?;
-            let value = tree
-                .lookup_entry_by_path(Path::new(path))
-                .map_err(gx)?
-                .map(|entry| (entry.object_id(), entry.mode().value()));
-            if self.db.is_some() {
-                self.dirty_paths.push((key, value.map(|(id, mode)| (id.to_string(), mode))));
-            }
-            value
-        };
-        if self.path_memory.len() >= 4096 {
-            self.flush();
-            self.path_memory.clear();
-        }
-        self.path_memory.insert(memory_key, value);
-        Ok(value)
     }
 }
 

crates/ide-vcs/src/repo.rs

@@ -65,7 +65,6 @@ pub(crate) fn discover(path: &Path) -> IdeResult<Option<Repo>> {
         inner: std::sync::Mutex::new(cache),
         ignored: Default::default(),
         log_index: Default::default(),
-        index_progress: Default::default(),
         log_signatures: Default::default(),
     }))
 }

crates/ide-vcs/tests/log_projection.rs

@@ -99,8 +99,6 @@ fn lazy_summaries_seek_and_index_survive_refresh_and_restart() {
     let cancel = ide_core::CancellationToken::new();
     cancel.cancel();
     assert!(repo.log_cancellable(&query, &cancel).is_err());
-    let progress = repo.index_log(&ide_core::CancellationToken::new()).unwrap();
-    assert_eq!((progress.indexed, progress.total, progress.running), (70, 70, false));
     assert_eq!(repo.log(&LogQuery { text: Some("needle body 5".into()), ..query.clone() }).unwrap().len(), 5);
     let new = commit(dir, "b", "new head");
     let refreshed = repo.log(&LogQuery { retain_until: Some(ids[0].clone()), ..query.clone() }).unwrap();
@@ -436,3 +434,31 @@ fn multiple_paths_are_simplified_together_at_merges() {
         assert_eq!(log[0].graph_parents.as_ref().unwrap().len(), 2);
     }
 }
+
+#[test]
+fn paths_are_literal_and_text_matches_bodies_or_hash_prefixes() {
+    let tmp = tempfile::tempdir().unwrap();
+    let dir = tmp.path();
+    git(dir, &["init", "-q"]);
+    let star = commit(dir, "[a]", "star\n\nBody Needle");
+    let plain = commit(dir, "a", "plain");
+    let repo = Repo::discover(dir).unwrap().unwrap();
+    let ids = |q: LogQuery| repo.log(&q).unwrap().into_iter().map(|c| c.id).collect::<Vec<_>>();
+    assert_eq!(ids(LogQuery { path: Some("[a]".into()), ..Default::default() }), vec![star.clone()]);
+    assert_eq!(ids(LogQuery { text: Some("body needle".into()), ..Default::default() }), vec![star.clone()]);
+    assert!(ids(LogQuery { text: Some("body needle".into()), text_match_case: true, ..Default::default() }).is_empty());
+    assert_eq!(ids(LogQuery { text: Some(plain[..7].to_uppercase()), ..Default::default() }), vec![plain]);
+}
+
+#[test]
+fn case_insensitive_text_folds_non_ascii() {
+    let tmp = tempfile::tempdir().unwrap();
+    let dir = tmp.path();
+    git(dir, &["init", "-q"]);
+    let hit = commit(dir, "a", "Ärger ПРИВЕТ");
+    commit(dir, "a", "other");
+    let repo = Repo::discover(dir).unwrap().unwrap();
+    let ids = |q: LogQuery| repo.log(&q).unwrap().into_iter().map(|c| c.id).collect::<Vec<_>>();
+    assert_eq!(ids(LogQuery { text: Some("ärger привет".into()), ..Default::default() }), vec![hit.clone()]);
+    assert!(ids(LogQuery { text: Some("ärger".into()), text_match_case: true, ..Default::default() }).is_empty());
+}

src-tauri/src/commands/mod.rs

@@ -262,7 +262,6 @@ pub fn builder() -> tauri_specta::Builder<tauri::Wry> {
         vcs::vcs_open_external_log,
         vcs::vcs_log_details,
         vcs::vcs_log_details_batch,
-        vcs::vcs_log_index,
         vcs::vcs_revision_children,
         vcs::vcs_revision_files,
         vcs::vcs_commit_files,

src-tauri/src/commands/vcs.rs

@@ -142,16 +142,6 @@ pub async fn vcs_log_details_batch(
 }
 #[tauri::command]
 #[specta::specta]
-pub async fn vcs_log_index(
-    window: Window,
-    state: State<'_, AppState>,
-    repo: Option<PathBuf>,
-    request: Option<RequestId>,
-) -> IdeResult<LogIndexProgress> {
-    ide_host::ops::vcs::vcs_log_index(&state.host, window.label(), repo, request).await
-}
-#[tauri::command]
-#[specta::specta]
 pub async fn vcs_revision_children(
     window: Window,
     state: State<'_, AppState>,

src/bindings.ts

@@ -496,7 +496,6 @@ export const commands = {
 	vcsOpenExternalLog: (path: string) => typedError<IpcPath, IdeError>(__TAURI_INVOKE("vcs_open_external_log", { path })),
 	vcsLogDetails: (repo: string | null, revision: string) => typedError<LogDetails, IdeError>(__TAURI_INVOKE("vcs_log_details", { repo, revision })),
 	vcsLogDetailsBatch: (repo: string | null, revisions: string[], request: RequestId) => typedError<LogDetails[], IdeError>(__TAURI_INVOKE("vcs_log_details_batch", { repo, revisions, request })),
-	vcsLogIndex: (repo: string | null, request: number | null) => typedError<LogIndexProgress, IdeError>(__TAURI_INVOKE("vcs_log_index", { repo, request })),
 	vcsRevisionChildren: (repo: string | null, revision: string, directory: string) => typedError<string[], IdeError>(__TAURI_INVOKE("vcs_revision_children", { repo, revision, directory })),
 	vcsRevisionFiles: (repo: string | null, revision: string) => typedError<string[], IdeError>(__TAURI_INVOKE("vcs_revision_files", { repo, revision })),
 	vcsCommitFiles: (repo: string | null, commit: string) => typedError<ChangedFile[], IdeError>(__TAURI_INVOKE("vcs_commit_files", { repo, commit })),
@@ -4987,12 +4986,6 @@ export type LogDetails = {
 	fingerprint: string,
 };
 
-export type LogIndexProgress = {
-	indexed: number,
-	total: number,
-	running: boolean,
-};
-
 export type LogLevel = "debug" | "info" | "warn" | "error";
 
 export type LogLocation = {

src/features/vcs/GitLog.tsx

@@ -49,7 +49,6 @@ import {
 import { hasPrimaryModifier } from "@/platform/keymap";
 import { laneColor, reachableFrom, type GraphRow, type GraphSegment } from "./graph";
 import { nextSpeedMatch } from "./logSearch";
-import { LogIndexStatus } from "./logIndex";
 import { presentLogGraph } from "./logPresentation";
 import {
   closeLogTab,
@@ -328,7 +327,6 @@ export function LogView() {
       <SplitPanel id="commits" defaultSize="54%" minSize="25%">
         <div className={styles.log}>
           <FilterBar />
-          <LogIndexStatus repositories={repo} />
           <CommitList />
         </div>
       </SplitPanel>

src/features/vcs/__tests__/logUiFeatures.test.tsx

@@ -23,7 +23,6 @@ vi.mock("@/platform/ipc", async (original) => {
         status: "ok",
         data: await Promise.all(ids.map(async () => (await api.details()).data)),
       }),
-      vcsLogIndex: () => ok({ indexed: 0, total: 0, running: false }),
       vcsDiffRevisions: (...args: unknown[]) => api.diff(...args),
     },
   };

src/features/vcs/logIndex.tsx

@@ -1,110 +0,0 @@
-import { useEffect } from "react";
-import { create } from "zustand";
-import type { LogIndexProgress } from "@/bindings";
-import { call, commands, nextRequestId, type RequestId } from "@/platform/ipc";
-import { tx } from "@/platform/i18n";
-import { useVcs } from "./store";
-import { IconButton } from "@/ui";
-
-interface Job {
-  version?: number;
-  restart?: boolean;
-  users: number;
-  request: RequestId | null;
-  paused: boolean;
-  progress: LogIndexProgress;
-  timer?: ReturnType<typeof setInterval>;
-}
-const jobs = new Map<string, Job>();
-const revision = create<{ value: number }>(() => ({ value: 0 }));
-const notify = () => revision.setState((s) => ({ value: s.value + 1 }));
-function start(repo: string, job: Job) {
-  if (job.paused) return;
-  if (job.request != null) {
-    job.restart = true;
-    return;
-  }
-  const request = nextRequestId();
-  job.request = request;
-  job.progress = { ...job.progress, running: true };
-  job.timer = setInterval(() => {
-    void call(commands.vcsLogIndex(repo, null))
-      .then((progress) => {
-        if (job.request === request) {
-          job.progress = progress;
-          notify();
-        }
-      })
-      .catch(() => undefined);
-  }, 750);
-  void call(commands.vcsLogIndex(repo, request))
-    .then(
-      (progress) => {
-        job.progress = progress;
-      },
-      () => {
-        job.progress = { ...job.progress, running: false };
-      },
-    )
-    .finally(() => {
-      if (job.request !== request) return;
-      job.request = null;
-      clearInterval(job.timer);
-      if (job.restart && job.users && !job.paused) {
-        job.restart = false;
-        start(repo, job);
-      }
-      notify();
-    });
-  notify();
-}
-
-/** Visible log scopes share one background job per repository. */
-export function LogIndexStatus({ repositories }: { repositories: readonly string[] }) {
-  revision((s) => s.value);
-  const version = useVcs((s) => s.version);
-  useEffect(() => {
-    for (const repo of repositories) {
-      const job: Job = jobs.get(repo) ?? {
-        users: 0,
-        request: null,
-        paused: false,
-        progress: { indexed: 0, total: 0, running: false },
-      };
-      jobs.set(repo, job);
-      const first = job.users++ === 0;
-      if (first || job.version !== version) {
-        job.version = version;
-        start(repo, job);
-      }
-    }
-    return () => {
-      for (const repo of repositories) {
-        const job = jobs.get(repo)!;
-        if (--job.users === 0) {
-          clearInterval(job.timer);
-          if (job.request != null) void commands.cancelRequest(job.request).catch(() => undefined);
-        }
-      }
-    };
-  }, [repositories, version]);
-  return repositories.map((repo) => {
-    const job = jobs.get(repo);
-    if (!job || (!job.paused && !job.progress.running)) return null;
-    return (
-      <div key={repo} role="status" style={{ display: "flex", alignItems: "center", gap: 6 }}>
-        {tx`Indexing Git log: ${job.progress.indexed} / ${job.progress.total}`}
-        <IconButton
-          icon={job.paused ? "play" : "pause"}
-          title={job.paused ? "Resume Indexing" : "Pause Indexing"}
-          onClick={() => {
-            job.paused = !job.paused;
-            if (job.paused && job.request != null) void commands.cancelRequest(job.request).catch(() => undefined);
-            else start(repo, job);
-            notify();
-          }}
-        />
-      </div>
-    );
-  });
-}

src/platform/locales/text/es.json

@@ -6164,9 +6164,6 @@
   "Shallow history boundary": "Límite de historial superficial (shallow)",
   "All Parents": "Todos los padres",
   "Show Commit Bookmark": "Mostrar marcador de confirmación",
-  "Indexing Git log: {indexed} / {total}": "Indexando registro de Git: {indexed} / {total}",
-  "Resume Indexing": "Reanudar indexación",
-  "Pause Indexing": "Pausar indexación",
   "Incompatible": "Incompatible",
   "Needs plugin API {engine}. Update OpenIDEA to install it.": "Requiere la API de plugins {engine}. Actualice OpenIDEA para instalarlo.",
   "Icon Themes": "Temas de iconos",

src/platform/locales/text/zh.json

@@ -6200,9 +6200,6 @@
   "Shallow history boundary": "浅克隆历史边界",
   "All Parents": "所有父提交",
   "Show Commit Bookmark": "显示提交书签",
-  "Indexing Git log: {indexed} / {total}": "正在索引 Git 日志：{indexed} / {total}",
-  "Resume Indexing": "继续索引",
-  "Pause Indexing": "暂停索引",
   "Incompatible": "不兼容",
   "Needs plugin API {engine}. Update OpenIDEA to install it.": "需要插件 API {engine}。请更新 OpenIDEA 后再安装。",
   "Icon Themes": "图标主题",