Fix spelling mistakes (#140)

Signed-off-by: Colin Ian King <[email protected]>

Author: ColinIanKing

CMakeLists.txt

@@ -47,7 +47,7 @@ target_include_directories(
 )
 
 # We use C++ 20 features in our exposed headers.
-# Anyone using us as a depdency and including our headers will need to be C++20 compatible.
+# Anyone using us as a dependency and including our headers will need to be C++20 compatible.
 #
 # Keep this variable in-sync with the `SVS_CXX_STANDARD` given below.
 # We need to manually set the standard for `spdlog`
@@ -148,7 +148,7 @@ configure_package_config_file(
     INSTALL_DESTINATION "${LIB_CONFIG_INSTALL_DIR}"
 )
 
-# Don't make compatibility guarentees until we reach a compatibility milestone.
+# Don't make compatibility guarantees until we reach a compatibility milestone.
 write_basic_package_version_file(
     ${VERSION_CONFIG}
     VERSION ${SVS_VERSION}

HISTORY.md

@@ -161,7 +161,7 @@ Finally, if environment variable based initialization is not desired, it can be
   **Reconstruction Semantics**
   * Uncompressed data is returned directly (potentially promoting to `float32`).
   * LVQ compressed data is reconstructed using this highest precision possible. For two
-    level datasets, boths levels will be used.
+    level datasets, both levels will be used.
   * LeanVec datasets will reconstruct using the full-precision secondary dataset.
 
 * Added an upgrade tool `pysvs.upgrader.upgrade` to upgrade the serialization layout of SVS
@@ -440,7 +440,7 @@ removal.
 * Version `v0.0.2` of serialized LVQ datasets is *broken*, the current version is now
   `v0.0.3`. This change was made to facilitate a canonical on-disk representation of LVQ.
 
-  Goind forward, previously saved LVQ formats can be reloaded using different runtime
+  Going forward, previously saved LVQ formats can be reloaded using different runtime
   alignments and different packing strategies without requiring whole dataset recompression.
 
   Any previously saved datasets will need to be regenerated from uncompressed data.

benchmark/include/svs-benchmark/benchmark.h

@@ -242,7 +242,7 @@ class ExecutableDispatcher {
 // new results as they become available and regularly saves results.
 //
 // It *does* involve many copies of the underlying TOML data, but I believe the tradoff in
-// data safety greatly outweights any extra time spent moving around TOML data.
+// data safety greatly outweighs any extra time spent moving around TOML data.
 class Checkpoint {
   private:
     toml::table data_{};
@@ -265,7 +265,7 @@ class Checkpoint {
         // Make sure to handle the case where this is the first data being registered
         // with the given key.
         //
-        // This is not the most efficient implemenation because we make unnecessary copies,
+        // This is not the most efficient implementation because we make unnecessary copies,
         // but the complexitry required to correctly applying incremental data does not
         // seem to be worth it.
         auto data_copy = data_;
@@ -304,7 +304,7 @@ struct BuildTime {
     }
 };
 
-/// Class shared to record the time taken to load an index into a useable form.
+/// Class shared to record the time taken to load an index into a usable form.
 struct LoadTime {
   public:
     double load_time_;

benchmark/include/svs-benchmark/executable.h

@@ -71,7 +71,7 @@ inline constexpr IsValidate is_validate{};
 ///     // Return an example job to serve as a prototype.
 ///     job_type example() const;
 ///
-///     // The name assoicated with the job. This will be used to both pull jobs from
+///     // The name associated with the job. This will be used to both pull jobs from
 ///     // the input TOML file as well as be the key under which results will be collected
 ///     // in output TOML file.
 ///     std::convertible_to<std::string> name() const;
@@ -100,7 +100,7 @@ inline constexpr IsValidate is_validate{};
 ///
 template <typename Implementation>
 class JobBasedExecutable : private Implementation, public Benchmark {
-    // Type Alises
+    // Type Aliases
   public:
     using job_type = typename Implementation::job_type;
     using dispatcher_type = typename Implementation::dispatcher_type;
@@ -254,13 +254,13 @@ class JobBasedExecutable : private Implementation, public Benchmark {
 ///
 ///     // Return a dispatcher for the executable.
 ///     // Each test must take `const job_type&` as an argument and return a destructurable
-///     // pair consiting of a string key and a `toml::table` of results.
+///     // pair consisting of a string key and a `toml::table` of results.
 ///     std::vector</*implementation-defined*/> tests() const;
 ///
 ///     // Return an example job to serve as a prototype.
 ///     job_type example() const;
 ///
-///     // The name assoicated with the job. This will be used to both pull jobs from
+///     // The name associated with the job. This will be used to both pull jobs from
 ///     // the input TOML file as well as be the key under which results will be collected
 ///  T   // in output TOML file.
 ///     std::convertible_to<std::string> name() const;
@@ -289,7 +289,7 @@ class JobBasedExecutable : private Implementation, public Benchmark {
 ///
 template <typename Implementation>
 class TestBasedExecutable : private Implementation, public Benchmark {
-    // Type Alises
+    // Type Aliases
   public:
     using job_type = typename Implementation::job_type;
 

benchmark/include/svs-benchmark/vamana/build.h

@@ -146,7 +146,7 @@ struct BuildJobBase {
     size_t num_threads_;
 
   public:
-    ///// Contructor
+    ///// Constructor
     BuildJobBase(
         std::string_view description,
         svsbenchmark::Dataset dataset,

benchmark/include/svs-benchmark/vamana/search.h

@@ -131,7 +131,7 @@ struct SearchJob {
         return search_parameters_;
     }
 
-    // Compatbility with `ExpectedResults`
+    // Compatibility with `ExpectedResults`
     static std::nullopt_t get_build_parameters() { return std::nullopt; }
     svs::DistanceType get_distance() const { return distance_; }
 

benchmark/src/vamana/build.cpp

@@ -102,7 +102,7 @@ Compiled specializations are listed below:
 
 template <typename BenchmarkType> struct Exe {
   public:
-    // type alises
+    // type aliases
     using job_type = associated_job_t<BenchmarkType>;
     using dispatcher_type = typename BuildDispatcher<BenchmarkType>::type;
 

bindings/python/microarch.py

@@ -50,7 +50,7 @@ def resolve_microarch(name: str):
         "native": cpu.host().name,
     }
     # Allow the custom aliases to override the current name.
-    # If an alias doesn't exist, juse pass the name straight through.
+    # If an alias doesn't exist, just pass the name straight through.
     return custom_aliases.get(name, name)
 
 def dump_flags_for_cmake(flags: list, path: str):

bindings/python/src/dynamic_vamana.cpp

@@ -154,7 +154,7 @@ the index monotonically increasing.
 const char* COMPACT_DOCSTRING = R"(
 Remove any holes created in the graph and data by renumbering internal IDs.
 Shrink the underlying data structures.
-Following ``consolidate``, this can potentialy reduce the memory footprint of the index
+Following ``consolidate``, this can potentially reduce the memory footprint of the index
 if a sufficient number of points were deleted.
 )";
 
@@ -264,7 +264,7 @@ void wrap(py::module& m) {
     add_threading_interface(vamana);
     add_data_interface(vamana);
 
-    // Vamana specific extentions.
+    // Vamana specific extensions.
     vamana::add_interface(vamana);
 
     // Dynamic interface.
@@ -358,7 +358,7 @@ auxiliary files that are needed when saving the various components of the index.
 
 If the directory does not exist, it will be created if its parent exists.
 
-It is the caller's responsibilty to ensure that no existing data will be
+It is the caller's responsibility to ensure that no existing data will be
 overwritten when saving the index to this directory.
     )"
     );

bindings/python/src/python_bindings.cpp

@@ -136,7 +136,7 @@ class ScopedModuleNameOverride {
 } // namespace
 
 PYBIND11_MODULE(_svs, m) {
-    // Internall, the top level `__init__.py` imports everything from the C++ module named
+    // Internally, the top level `__init__.py` imports everything from the C++ module named
     // `_svs`.
     //
     // Performing the name override makes the definitions inside the C++ bindings

bindings/python/src/svs/common.py

@@ -133,7 +133,7 @@ def read_svs(filename: str, dtype = np.float32):
     with open(filename, "rb") as fin:
         # Read through the magic number
         struct.unpack('q', fin.read(8))
-        # Read throug the UUID
+        # Read through the UUID
         _uuid = struct.unpack('q', fin.read(8))
         _uuid = struct.unpack('q', fin.read(8))
 
@@ -159,7 +159,7 @@ def write_vecs(array: np.array, filename: str, skip_check: bool = False):
     Result:
         The array is saved to the requested file.
 
-    File extention to array element type:
+    File extension to array element type:
 
         * fvecs: np.float32
         * hvecs: np.float16
@@ -201,7 +201,7 @@ def write_vecs(array: np.array, filename: str, skip_check: bool = False):
             io.write(array[i,:].tobytes())
 
 def random_dataset(nvectors: int, ndims: int, dtype = np.float32, seed = None):
-    # Use the `RandomState` generator for it's strong backward-compatibility guarentee.
+    # Use the `RandomState` generator for it's strong backward-compatibility guarantee.
     # I.E., we don't care about performance. We want reproducibility here.
     rng = np.random.RandomState(seed = seed)
     if dtype in (np.float16, np.float32):

bindings/python/src/vamana.cpp

@@ -541,9 +541,9 @@ auxiliary files that are needed when saving the various components of the index.
 
 If the directory does not exist, it will be created if its parent exists.
 
-It is the caller's responsibilty to ensure that no existing data will be
+It is the caller's responsibility to ensure that no existing data will be
 overwritten when saving the index to this directory.
     )"
     );
 }
-} // namespace svs::python::vamana
+} // namespace svs::python::vamana

bindings/python/tests/dynamic.py

@@ -72,7 +72,7 @@ def ids(self):
         return self.current_ids
 
     def ground_truth(self, num_neighbors: int):
-        # Gather the dataset into a contiguous chunck to pass to the ground truth
+        # Gather the dataset into a contiguous chunk to pass to the ground truth
         # calculation.
         ids_np = np.array(list(self.current_ids), dtype = np.uint64)
         sub_dataset = self.raw_data[ids_np, :]

bindings/python/tests/test_dynamic_vamana.py

@@ -76,7 +76,7 @@ def recall_check(
             I, D = reloaded.search(reference.queries, num_neighbors)
             reloaded_recall = svs.k_recall_at(gt, I, num_neighbors, num_neighbors)
 
-            # Because saving triggers graph compaction, we can't guarentee that the reloaded
+            # Because saving triggers graph compaction, we can't guarantee that the reloaded
             # recall is the same as the original index.
             print(f"    Reloaded Recall: {reloaded_recall}")
             self.assertTrue(reloaded_recall < expected_recall + recall_delta)

bindings/python/tests/test_flat.py

@@ -102,7 +102,7 @@ def _do_test(self, flat, queries, groundtruth, distance, data = svs.read_vecs(te
         )
 
         # Test querying.
-        # Return as many neighbors as we have exising groundtruth for.
+        # Return as many neighbors as we have existing groundtruth for.
         num_neighbors = groundtruth.shape[-1]
         results = flat.search(queries, num_neighbors)
 

examples/cpp/custom_thread_pool.cpp

@@ -278,6 +278,6 @@ int svs_main(std::vector<std::string> args) {
     return 0;
 }
 
-// Special main providing some helpful utilties.
+// Special main providing some helpful utilities.
 SVS_DEFINE_MAIN();
 //! [Example All]

examples/cpp/saveload.cpp

@@ -57,7 +57,7 @@ class ContextFreeSaveable {
     // Serialized objects need a schema as well, which is essentially a unique name
     // associated with the serialized TOML table.
     //
-    // The combination of schema and name allow speculative loading code some guarentee
+    // The combination of schema and name allow speculative loading code some guarantee
     // as to the expected contents and types of a table.
     static constexpr std::string_view serialization_schema = "example_context_free";
 

examples/cpp/shared/shared.cpp

@@ -136,5 +136,5 @@ int svs_main(std::vector<std::string> SVS_UNUSED(args)) {
     return 0;
 }
 
-// Special main providing some helpful utilties.
+// Special main providing some helpful utilities.
 SVS_DEFINE_MAIN();

examples/cpp/vamana.cpp

@@ -191,6 +191,6 @@ int svs_main(std::vector<std::string> args) {
     return 0;
 }
 
-// Special main providing some helpful utilties.
+// Special main providing some helpful utilities.
 SVS_DEFINE_MAIN();
 //! [Example All]

include/svs/concepts/data.h

@@ -61,14 +61,14 @@ namespace data {
 ///    // Require that ``T`` has the type alias ``T::value_type``.
 ///    //
 ///    // Note that the the alias does not necessarily need to be a "value_type" in the
-///    // sence of C++ value type. In other words, it can (and probably should be for
+///    // sense of C++ value type. In other words, it can (and probably should be for
 ///    // performance reasons be a reference).
 ///    typename T::value_type;
 ///
 ///    // Require that ``T`` has the type alias ``T::const_value_type``.
 ///    //
 ///    // Note that the the alias does not necessarily need to be a "value_type" in the
-///    // sence of C++ value type. In other words, it can (and probably should be for
+///    // sense of C++ value type. In other words, it can (and probably should be for
 ///    // performance reasons be a reference).
 ///    typename T::const_value_type;
 /// };

include/svs/core/allocator.h

@@ -18,7 +18,7 @@
 
 ///
 /// @file
-/// @brief Implements commont large-scale allocators used by the many data structures.
+/// @brief Implements common large-scale allocators used by the many data structures.
 ///
 
 ///
@@ -44,7 +44,7 @@
 
 // <linux/mman.h> provides some linux-specific flags like
 // MAP_POPULATE, MAP_NORESERVE, MAP_HUGETLB.
-// They can be used only with linux, and are not availiable
+// They can be used only with linux, and are not available
 // for MacOS
 
 #if defined(__linux__)

include/svs/core/data.h

@@ -91,7 +91,7 @@ struct UnspecializedVectorDataLoader {
 ///
 /// @tparam T The element type of the encoded vectors.
 ///     Typically, this will be a floating point type like ``float`` or ``svs::Float16``
-///     but may be an integer type as well for certain datsets.
+///     but may be an integer type as well for certain datasets.
 /// @tparam Extent The compile-time dimensionality of the vectors to be read.
 ///     May provide a performance boost if given. Default: ``svs::Dynamic``.
 /// @tparam Allocator The allocator to use for the memory backing the data when loaded.

include/svs/core/data/io.h

@@ -31,7 +31,7 @@ namespace svs::io {
 // Dispatch tags to control the loading and saving pipeline.
 constexpr lib::PriorityTag<2> default_populate_tag = lib::PriorityTag<2>();
 
-// Some specialized datasets may need to perform aritrary computation
+// Some specialized datasets may need to perform arbitrary computation
 // or rearrangement of the saved data prior to committing it to the dataset.
 //
 // These datasets are expected to provide their own custom accessor in this case.
@@ -187,7 +187,7 @@ inline bool special_by_file_extension(std::string_view path) {
 /// the dataset and (2) the number of dimensions for each vector and return an allocated
 /// dataset capable of holding a dataset with those dimensions.
 ///
-/// Recognized file extentions:
+/// Recognized file extensions:
 /// * .svs: The native file format for this library.
 /// * .vecs: The usual [f/b/i]vecs form.
 /// * .bin: Files generated by DiskANN.

include/svs/core/data/simple.h

@@ -150,7 +150,7 @@ struct Matcher {
     }
 
     // Don't support deduction on the try-load path.
-    // Deduction throws too many exceptions to handle corectly right now.
+    // Deduction throws too many exceptions to handle correctly right now.
     static lib::TryLoadResult<Matcher> try_load_direct(
         const std::filesystem::path& SVS_UNUSED(path),
         svs::DataType SVS_UNUSED(type_hint) = svs::DataType::undef,
@@ -478,7 +478,7 @@ class SimpleData {
     void compact(
         std::span<const I> new_to_old, Pool& threadpool, size_t batchsize = 1'000'000
     ) {
-        // Alllocate scratch space.
+        // Allocate scratch space.
         batchsize = std::min(batchsize, size());
         auto buffer = data::SimpleData<T, Extent>(batchsize, dimensions());
         compact_data(*this, buffer, new_to_old, threadpool);
@@ -771,7 +771,7 @@ class SimpleData<T, Extent, Blocked<Alloc>> {
     template <std::integral I, threads::ThreadPool Pool>
     void
     compact(std::span<const I> new_to_old, Pool& threadpool, size_t batchsize = 1'000'000) {
-        // Alllocate scratch space.
+        // Allocate scratch space.
         batchsize = std::min(batchsize, size());
         auto buffer = data::SimpleData<T, Extent>(batchsize, dimensions());
         compact_data(*this, buffer, new_to_old, threadpool);

include/svs/core/distance.h

@@ -155,7 +155,7 @@ class DistanceDispatcher {
     ///     to construct the dispatcher.
     ///
     ///     All other arguments will be forwarded to ``f`` beginning at argument position 2.
-    /// @param args Arguements to forward to ``f``.
+    /// @param args Arguments to forward to ``f``.
     ///
     template <typename F, typename... Args> auto operator()(F&& f, Args&&... args) {
         switch (distance_type_) {

include/svs/core/distance/cosine.h

@@ -53,7 +53,7 @@ class CosineSimilarity {
             );
         }
         if (__builtin_expect(svs::detail::avx_runtime_flags.is_avx2_supported(), 1)) {
-            // We do not support AVX2 on CS yet, but it will fallabck to generic anyway
+            // We do not support AVX2 on CS yet, but it will fallback to generic anyway
             return CosineSimilarityImpl<Dynamic, Ea, Eb, AVX_AVAILABILITY::AVX2>::compute(
                 a, b, a_norm, lib::MaybeStatic(N)
             );
@@ -146,7 +146,7 @@ inline constexpr bool operator==(DistanceCosineSimilarity, DistanceCosineSimilar
 ///
 /// @ingroup distance_overload
 /// @anchor compute_distancecosine
-/// @brief Compute the Cosine simmilarity between two vectors in R^n.
+/// @brief Compute the Cosine similarity between two vectors in R^n.
 ///
 /// @tparam Ea The element type for each component of the left-hand argument.
 /// @tparam Eb The element type for each component of the right-hand argument.