Lint: Use … instead of ... (rerun-io#3100)

### What
The ellipsis character `…` should be used in GUIs and docs instead of
the similar-looking `...`. The former looks better over all, and the
latter can have a linebreak inserted in the middle of it which doesn't
look great. It is also for consistency.

Ideally we would only lint strings in Rust using something like
https://github.com/trailofbits/dylint but I'll leave that as an exercise
for later 😬

For our developers: I suggest you figure out how to type `…` quickly on
your keyboard. On my Mac I press `⌥;` (but I can't remember if that is
something special I've configured…)

This one is for you @martenbjork 

### Checklist
* [x] I have read and agree to [Contributor
Guide](https://github.com/rerun-io/rerun/blob/main/CONTRIBUTING.md) and
the [Code of
Conduct](https://github.com/rerun-io/rerun/blob/main/CODE_OF_CONDUCT.md)
* [x] I've included a screenshot or gif (if applicable)
* [x] I have tested [demo.rerun.io](https://demo.rerun.io/pr/3100) (if
applicable)

- [PR Build Summary](https://build.rerun.io/pr/3100)
- [Docs
preview](https://rerun.io/preview/9de65e45790b192899248abe12d1a7b43b1ca488/docs)
<!--DOCS-PREVIEW-->
- [Examples
preview](https://rerun.io/preview/9de65e45790b192899248abe12d1a7b43b1ca488/examples)
<!--EXAMPLES-PREVIEW--><!--EXAMPLES-PREVIEW-->
- [Recent benchmark results](https://ref.rerun.io/dev/bench/)
- [Wasm size tracking](https://ref.rerun.io/dev/sizes/)

README.md

@@ -87,5 +87,5 @@ The commercial product targets the needs specific to teams that build and run co
 ## Installing a pre-release Python SDK
 
 1. Download the correct `.whl` from [GitHub Releases](https://github.com/rerun-io/rerun/releases)
-2. Run `pip install rerun_sdk<...>.whl` (replace `<...>` with the actual filename)
+2. Run `pip install rerun_sdk<…>.whl` (replace `<…>` with the actual filename)
 3. Test it: `rerun --version`

crates/re_analytics/src/pipeline_native.rs

@@ -61,7 +61,7 @@ impl Pipeline {
         //
         // Joining these threads is not a viable strategy for two reasons:
         // 1. We _never_ want to delay the shutdown process, analytics must never be in the way.
-        // 2. We need to deal with unexpected shutdowns anyway (crashes, SIGINT, SIGKILL, ...),
+        // 2. We need to deal with unexpected shutdowns anyway (crashes, SIGINT, SIGKILL, …),
         //    and we do indeed.
         //
         // This is an at-least-once pipeline: in the worst case, unexpected shutdowns will lead to
@@ -213,7 +213,7 @@ fn realtime_pipeline(
             return;
         }
         if let Err(err) = session_file.rewind() {
-            // We couldn't reset the session file... That one is a bit messy and will likely break
+            // We couldn't reset the session file… That one is a bit messy and will likely break
             // analytics for the entire duration of this session, but that really _really_ should
             // never happen.
             re_log::debug_once!("couldn't seek into analytics data file: {err}");
@@ -223,7 +223,7 @@ fn realtime_pipeline(
     let on_event = |session_file: &mut _, event| {
         re_log::trace!(
             %analytics_id, %session_id,
-            "appending event to current session file..."
+            "appending event to current session file…"
         );
         if let Err(event) = append_event(session_file, &analytics_id, &session_id, event) {
             // We failed to append the event to the current session, so push it back at the end of

crates/re_arrow_store/src/store.rs

@@ -133,7 +133,7 @@ impl<T: Clone> std::ops::DerefMut for MetadataRegistry<T> {
     }
 }
 
-/// Used to cache auto-generated cluster cells (`[0]`, `[0, 1]`, `[0, 1, 2]`, ...) so that they
+/// Used to cache auto-generated cluster cells (`[0]`, `[0, 1]`, `[0, 1, 2]`, …) so that they
 /// can be properly deduplicated on insertion.
 #[derive(Debug, Default, Clone)]
 pub struct ClusterCellCache(pub IntMap<u32, DataCell>);
@@ -200,7 +200,7 @@ pub struct DataStore {
     /// Only used to map `RowId`s to their original [`TimePoint`]s at the moment.
     pub(crate) metadata_registry: MetadataRegistry<TimePoint>,
 
-    /// Used to cache auto-generated cluster cells (`[0]`, `[0, 1]`, `[0, 1, 2]`, ...)
+    /// Used to cache auto-generated cluster cells (`[0]`, `[0, 1]`, `[0, 1, 2]`, …)
     /// so that they can be properly deduplicated on insertion.
     pub(crate) cluster_cell_cache: ClusterCellCache,
 
@@ -409,7 +409,7 @@ pub struct IndexedTable {
     /// buckets, in bytes.
     ///
     /// This is a best-effort approximation, adequate for most purposes (stats,
-    /// triggering GCs, ...).
+    /// triggering GCs, …).
     pub buckets_size_bytes: u64,
 }
 
@@ -506,7 +506,7 @@ pub struct IndexedBucketInner {
     /// included, in bytes.
     ///
     /// This is a best-effort approximation, adequate for most purposes (stats,
-    /// triggering GCs, ...).
+    /// triggering GCs, …).
     ///
     /// We cache this because there can be many, many buckets.
     pub size_bytes: u64,

crates/re_arrow_store/src/store_polars.rs

@@ -300,7 +300,7 @@ fn column_as_series(
         ListArray::<i32>::default_datatype(datatype.clone()),
         Offsets::try_from_lengths(comp_lengths).unwrap().into(),
         // It's possible that all rows being referenced were already garbage collected (or simply
-        // never existed to begin with), at which point `comp_rows` will be empty... and you can't
+        // never existed to begin with), at which point `comp_rows` will be empty… and you can't
         // call `concatenate` on an empty list without panicking.
         if comp_values.is_empty() {
             new_empty_array(datatype)

crates/re_arrow_store/src/store_read.rs

@@ -104,7 +104,7 @@ impl DataStore {
             id = self.query_id.load(Ordering::Relaxed),
             timeline = ?timeline,
             entity = %ent_path,
-            "query started..."
+            "query started…"
         );
 
         let timeless: Option<IntSet<_>> = self
@@ -215,7 +215,7 @@ impl DataStore {
             entity = %ent_path,
             %primary,
             ?components,
-            "query started..."
+            "query started…"
         );
 
         let cells = self
@@ -405,7 +405,7 @@ impl DataStore {
             query = ?query,
             entity = %ent_path,
             ?components,
-            "query started..."
+            "query started…"
         );
 
         let temporal = self
@@ -694,14 +694,14 @@ impl IndexedBucket {
             ?components,
             timeline = %self.timeline.name(),
             time = self.timeline.typ().format(time),
-            "searching for primary & secondary cells..."
+            "searching for primary & secondary cells…"
         );
 
         let time_row_nr = col_time.partition_point(|t| *t <= time.as_i64()) as i64;
 
         // The partition point is always _beyond_ the index that we're looking for.
         // A partition point of 0 thus means that we're trying to query for data that lives
-        // _before_ the beginning of time... there's nothing to be found there.
+        // _before_ the beginning of time… there's nothing to be found there.
         if time_row_nr == 0 {
             return None;
         }
@@ -818,7 +818,7 @@ impl IndexedBucket {
             ?components,
             timeline = %self.timeline.name(),
             time_range = self.timeline.typ().format_range(time_range),
-            "searching for time & component cell numbers..."
+            "searching for time & component cell numbers…"
         );
 
         let time_row_nr = col_time.partition_point(|t| *t < time_range.min.as_i64()) as u64;
@@ -1005,7 +1005,7 @@ impl PersistentIndexedTable {
             %primary,
             ?components,
             timeless = true,
-            "searching for primary & secondary cells..."
+            "searching for primary & secondary cells…"
         );
 
         // find the primary row number's row.

crates/re_arrow_store/src/store_stats.rs

@@ -368,7 +368,7 @@ impl IndexedBucketInner {
     /// stack and heap included, in bytes.
     ///
     /// This is a best-effort approximation, adequate for most purposes (stats,
-    /// triggering GCs, ...).
+    /// triggering GCs, …).
     #[inline]
     pub fn compute_size_bytes(&mut self) -> u64 {
         re_tracing::profile_function!();

crates/re_arrow_store/src/store_write.rs

@@ -129,7 +129,7 @@ impl DataStore {
                 .collect::<Vec<_>>(),
             entity = %ent_path,
             components = ?cells.iter().map(|cell| cell.component_name()).collect_vec(),
-            "insertion started..."
+            "insertion started…"
         );
 
         let cluster_cell_pos = cells
@@ -155,7 +155,7 @@ impl DataStore {
             None
         } else {
             // The caller has not specified any cluster component, and so we'll have to generate
-            // one... unless we've already generated one of this exact length in the past,
+            // one… unless we've already generated one of this exact length in the past,
             // in which case we can simply re-use that cell.
 
             Some(self.generate_cluster_cell(num_instances))

crates/re_arrow_store/tests/internals.rs

@@ -18,7 +18,7 @@ use re_types::{components::InstanceKey, Loggable as _};
 // is performance: getting the datastore into a pathological topology should show up in
 // integration query benchmarks.
 //
-// In the current state of things, though, it is much easier to test for it that way... so we
+// In the current state of things, though, it is much easier to test for it that way… so we
 // make an exception, for now...
 #[test]
 fn pathological_bucket_topology() {

crates/re_build_info/src/build_info.rs

@@ -7,7 +7,7 @@
 ///
 /// There are a few other cases though, like
 /// - `git` is not installed
-/// - the user downloaded rerun as a tarball and then imported via a `path = ...` import
+/// - the user downloaded rerun as a tarball and then imported via a `path = …` import
 /// - others?
 #[derive(Copy, Clone, Debug)]
 pub struct BuildInfo {

crates/re_build_tools/src/lib.rs

@@ -36,9 +36,9 @@ pub use self::rebuild_detector::{
 //
 // When working within the workspace, we can simply try and call `git` and we're done.
 //
-// # Using an unpublished crate (e.g. `path = "..."` or `git = "..."` or `[patch.crates-io]`)
+// # Using an unpublished crate (e.g. `path = "…"` or `git = "…"` or `[patch.crates-io]`)
 //
-// In these cases we may or may not have access to the workspace (e.g. a `path = ...` import likely
+// In these cases we may or may not have access to the workspace (e.g. a `path = …` import likely
 // will, while a crate patch won't).
 //
 // This is not an issue however, as we can simply try and see what we get.

crates/re_components/src/tensor.rs

@@ -438,8 +438,8 @@ impl Tensor {
 
         match shape_short.len() {
             1 => {
-                // Special case: Nx1(x1x1x...) tensors are treated as Nx1 gray images.
-                // Special case: Nx1(x1x1x...) tensors are treated as Nx1 gray images.
+                // Special case: Nx1(x1x1x…) tensors are treated as Nx1 gray images.
+                // Special case: Nx1(x1x1x…) tensors are treated as Nx1 gray images.
                 if self.shape.len() >= 2 {
                     Some([shape_short[0].size, 1, 1])
                 } else {

crates/re_data_store/src/store_db.rs

@@ -102,7 +102,7 @@ impl EntityDb {
                     DataCell::from_arrow_empty(cell.component_name(), cell.datatype().clone());
 
                 // NOTE(cmc): The fact that this inserts data to multiple entity paths using a
-                // single `RowId` is... interesting. Keep it in mind.
+                // single `RowId` is… interesting. Keep it in mind.
                 let row = DataRow::from_cells1(
                     row_id,
                     row.entity_path.clone(),
@@ -134,7 +134,7 @@ impl EntityDb {
                     DataCell::from_arrow_empty(component_path.component_name, data_type.clone());
 
                 // NOTE(cmc): The fact that this inserts data to multiple entity paths using a
-                // single `RowId` is... interesting. Keep it in mind.
+                // single `RowId` is… interesting. Keep it in mind.
                 let row = DataRow::from_cells1(
                     row_id,
                     component_path.entity_path.clone(),

crates/re_log_encoding/src/stream_rrd_from_http.rs

@@ -112,7 +112,7 @@ mod web_event_listener {
     ///
     /// From javascript you can send an rrd using:
     /// ``` ignore
-    /// var rrd = new Uint8Array(...); // Get an RRD from somewhere
+    /// var rrd = new Uint8Array(…); // Get an RRD from somewhere
     /// window.postMessage(rrd, "*")
     /// ```
     pub fn stream_rrd_from_event_listener(on_msg: Arc<HttpMessageCallback>) {

crates/re_log_types/src/data_cell.rs

@@ -42,7 +42,7 @@ pub type DataCellResult<T> = ::std::result::Result<T, DataCellError>;
 ///
 /// ## Layout
 ///
-/// A cell is an array of component instances: `[C, C, C, ...]`.
+/// A cell is an array of component instances: `[C, C, C, …]`.
 ///
 /// Consider this example:
 /// ```ignore
@@ -133,7 +133,7 @@ pub struct DataCellInner {
     /// costly operation.
     pub(crate) size_bytes: u64,
 
-    /// A uniformly typed list of values for the given component type: `[C, C, C, ...]`
+    /// A uniformly typed list of values for the given component type: `[C, C, C, …]`
     ///
     /// Includes the data, its schema and probably soon the component metadata
     /// (e.g. the `ComponentName`).
@@ -332,12 +332,12 @@ impl DataCell {
     ///
     /// Useful when dealing with cells of different lengths in context that don't allow for it.
     ///
-    /// * Before: `[C, C, C, ...]`
+    /// * Before: `[C, C, C, …]`
     /// * After: `ListArray[ [C, C, C, C] ]`
     //
     // TODO(#1696): this shouldn't be public, need to make it private once the store has been
     // patched to use datacells directly.
-    // TODO(cmc): effectively, this returns a `DataColumn`... think about that.
+    // TODO(cmc): effectively, this returns a `DataColumn`… think about that.
     #[doc(hidden)]
     #[inline]
     pub fn to_arrow_monolist(&self) -> Box<dyn arrow2::array::Array> {

crates/re_log_types/src/data_row.rs

@@ -171,7 +171,7 @@ impl std::ops::DerefMut for RowId {
 /// ## Layout
 ///
 /// A row is a collection of cells where each cell must either be empty (a clear), unit-lengthed
-/// (a splat) or `num_instances` long (standard): `[[C1, C1, C1], [], [C3], [C4, C4, C4], ...]`.
+/// (a splat) or `num_instances` long (standard): `[[C1, C1, C1], [], [C3], [C4, C4, C4], …]`.
 ///
 /// Consider this example:
 /// ```ignore

crates/re_log_types/src/data_table.rs

@@ -207,9 +207,9 @@ impl std::ops::DerefMut for TableId {
 /// (standard):
 /// ```text
 /// [
-///   [[C1, C1, C1], [], [C3], [C4, C4, C4], ...],
-///   [None, [C2, C2], [], [C4], ...],
-///   [None, [C2, C2], [], None, ...],
+///   [[C1, C1, C1], [], [C3], [C4, C4, C4], …],
+///   [None, [C2, C2], [], [C4], …],
+///   [None, [C2, C2], [], None, …],
 ///   ...
 /// ]
 /// ```
@@ -698,8 +698,8 @@ impl DataTable {
 
         /// Transforms an array of unit values into a list of unit arrays.
         ///
-        /// * Before: `[C, C, C, C, C, ...]`
-        /// * After: `ListArray[ [C], [C], [C], [C], [C], ... ]`
+        /// * Before: `[C, C, C, C, C, …]`
+        /// * After: `ListArray[ [C], [C], [C], [C], [C], … ]`
         // NOTE: keeping that one around, just in case.
         #[allow(dead_code)]
         fn unit_values_to_unit_lists(array: Box<dyn Array>) -> Box<dyn Array> {
@@ -795,8 +795,8 @@ impl DataTable {
 
         /// Create a list-array out of a flattened array of cell values.
         ///
-        /// * Before: `[C, C, C, C, C, C, C, ...]`
-        /// * After: `ListArray[ [[C, C], [C, C, C], None, [C], [C], ...] ]`
+        /// * Before: `[C, C, C, C, C, C, C, …]`
+        /// * After: `ListArray[ [[C, C], [C, C, C], None, [C], [C], …] ]`
         fn data_to_lists(
             column: &[Option<DataCell>],
             data: Box<dyn Array>,

crates/re_log_types/src/data_table_batcher.rs

@@ -418,7 +418,7 @@ fn batching_thread(
         // table.sort();
 
         // NOTE: This can only fail if all receivers have been dropped, which simply cannot happen
-        // as long the batching thread is alive... which is where we currently are.
+        // as long the batching thread is alive… which is where we currently are.
         tx_table.send(table).ok();
 
         acc.reset();