Remove dead pub code in diskann-tools and diskann-disk

diskann-disk/src/build/configuration/filter_parameter.rs

@@ -5,13 +5,6 @@
 
 use crate::data_model::GraphDataType;
 
-pub type AssociatedDataFilter<Data> =
-    Box<dyn Fn(&<Data as GraphDataType>::AssociatedDataType) -> bool>;
-
-pub fn default_associated_data_filter<Data: GraphDataType>() -> AssociatedDataFilter<Data> {
-    Box::new(|_| true)
-}
-
 pub type VectorFilter<'a, Data> =
     Box<dyn Fn(&<Data as GraphDataType>::VectorIdType) -> bool + Send + Sync + 'a>;
 
@@ -26,14 +19,6 @@ mod tests {
 
     type TestGraphData = GraphDataF32VectorUnitData;
 
-    #[test]
-    fn test_default_associated_data_filter_returns_true_for_all() {
-        let filter = default_associated_data_filter::<TestGraphData>();
-        // Test that the default filter always returns true
-        assert!(filter(&()));
-        assert!(filter(&()));
-    }
-
     #[test]
     fn test_default_vector_filter_returns_true_for_all() {
         let filter = default_vector_filter::<TestGraphData>();

diskann-tools/src/utils/mod.rs

@@ -24,12 +24,6 @@ pub use ground_truth::*;
 // pub mod range_search_disk_index;
 // pub use range_search_disk_index::*;
 
-pub mod search_disk_index;
-pub use search_disk_index::*;
-
-pub mod build_disk_index;
-pub use build_disk_index::*;
-
 pub mod build_pq;
 pub use build_pq::*;
 
@@ -39,9 +33,6 @@ pub use generate_synthetic_labels_utils::*;
 pub mod gen_associated_data_from_range;
 pub use gen_associated_data_from_range::*;
 
-pub mod test_utils;
-pub use test_utils::*;
-
 pub type CMDResult<T> = Result<T, CMDToolError>;
 
 pub mod parameter_helper;

diskann-tools/src/utils/random_data_generator.rs

@@ -197,7 +197,28 @@ mod tests {
     use rstest::rstest;
 
     use super::*;
-    use crate::utils::size_constants::{TEST_DATASET_SIZE_SMALL, TEST_NUM_DIMENSIONS_RECOMMENDED};
+
+    /// The small dataset size for testing the library.
+    /// A prime number is used to avoid any accidental patterns in the data.
+    #[cfg(not(miri))]
+    const TEST_DATASET_SIZE_SMALL: u64 = 101;
+
+    /// The recommended number of dimensions for testing the library.
+    /// A prime number is used to avoid any accidental patterns in the data.
+    /// When "memory aligned" the dimensions become 64 (8*8). Setting to non-aligned value to ensure aligning works.
+    #[cfg(not(miri))]
+    const TEST_NUM_DIMENSIONS_RECOMMENDED: usize = 59;
+
+    /// The small dataset size for testing the library.
+    /// A prime number is used to avoid any accidental patterns in the data.
+    #[cfg(miri)]
+    const TEST_DATASET_SIZE_SMALL: u64 = 3;
+
+    /// The recommended number of dimensions for testing the library.
+    /// A prime number is used to avoid any accidental patterns in the data.
+    /// When "memory aligned" the dimensions become 16 (8*2). Setting to non-aligned value to ensure aligning works.
+    #[cfg(miri)]
+    const TEST_NUM_DIMENSIONS_RECOMMENDED: usize = 13;
 
     #[rstest]
     fn random_data_write_success(