bump rust toolchain

.github/actions/build_linux/action.yml

@@ -70,9 +70,9 @@ runs:
         artifacts="meta,metactl,query,sqllogictests"
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact ..."
-          cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --bin databend-$artifact
+          cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --bin databend-$artifact
         done
-        cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --bin open-sharing
+        cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --bin open-sharing
         ls -lh ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-*
 
     - name: Build Debug for specific artifacts
@@ -82,7 +82,7 @@ runs:
         artifacts="${{ inputs.artifacts }}"
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact ..."
-          cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --bin databend-$artifact
+          cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --bin databend-$artifact
         done
         ls -lh ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-$artifact
 
@@ -93,7 +93,7 @@ runs:
         artifacts="meta,metactl,query,sqllogictests"
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact ..."
-          cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --release --bin databend-$artifact
+          cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --release --bin databend-$artifact
         done
         ls -lh ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-$artifact
 
@@ -104,7 +104,7 @@ runs:
         artifacts="${{ inputs.artifacts }}"
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact ..."
-          cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --release --bin databend-$artifact
+          cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --release --bin databend-$artifact
         done
         ls -lh ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-$artifact
 

.github/actions/build_linux_sanitizer/action.yml

@@ -53,7 +53,7 @@ runs:
         artifacts="meta,metactl,query"
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact with sanitizer ..."
-          cargo -Zbuild-std -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --manifest-path src/binaries/Cargo.toml --bin databend-$artifact
+          cargo -Zbuild-std -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --features ${{ inputs.features }} --manifest-path src/binaries/Cargo.toml --bin databend-$artifact
         done
         ls -lh ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-*
 

.github/actions/build_macos/action.yml

@@ -44,18 +44,18 @@ runs:
     - name: Build Debug for all artifacts
       if: env.BUILD_PROFILE == 'debug' && inputs.artifacts == 'all'
       shell: bash
-      run: cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }}
+      run: cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }}
 
     - name: Build Debug for specific artifacts
       if: env.BUILD_PROFILE == 'debug' && inputs.artifacts != 'all'
       shell: bash
       run: |
         artifacts="${{ inputs.artifacts }}"
         echo "==> building libs ..."
-        cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --lib
+        cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --lib
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact ..."
-          cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --bin databend-$artifact
+          cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --bin databend-$artifact
         done
 
     - name: Build Release
@@ -64,10 +64,10 @@ runs:
       run: |
         artifacts="${{ inputs.artifacts }}"
         echo "==> building libs ..."
-        cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --release --lib
+        cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --release --lib
         for artifact in ${artifacts//,/ }; do
           echo "==> building databend-$artifact ..."
-          cargo -Zgitoxide=fetch,shallow-index,shallow-deps build --target ${{ inputs.target }} --release --bin databend-$artifact
+          cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps build --target ${{ inputs.target }} --release --bin databend-$artifact
         done
 
     - name: Upload artifact

.github/actions/check/action.yml

@@ -47,4 +47,4 @@ runs:
 
     - name: Clippy
       shell: bash
-      run: cargo -Zgitoxide=fetch,shallow-index,shallow-deps clippy --workspace --all-targets --all-features -- -D warnings
+      run: cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps clippy --workspace --all-targets --all-features -- -D warnings

.github/actions/test_unit/action.yml

@@ -11,7 +11,7 @@ runs:
 
     - shell: bash
       run: |
-        cargo -Zgitoxide=fetch,shallow-index,shallow-deps nextest run --no-fail-fast --hide-progress-bar
+        cargo -Zgitoxide=fetch -Zgit=shallow-index,shallow-deps nextest run --no-fail-fast --hide-progress-bar
       env:
         RUST_TEST_THREADS: "8"
         RUST_LOG: ERROR

rust-toolchain.toml

@@ -1,3 +1,3 @@
 [toolchain]
-channel = "nightly-2023-10-23"
+channel = "nightly-2024-01-23"
 components = ["rustfmt", "clippy", "rust-src", "miri", "rust-analyzer"]

src/common/arrow/src/arrow/array/boolean/mod.rs

@@ -33,7 +33,6 @@ mod from;
 mod iterator;
 mod mutable;
 
-pub use iterator::*;
 pub use mutable::*;
 
 /// A [`BooleanArray`] is Arrow's semantically equivalent of an immutable `Vec<Option<bool>>`.

src/common/arrow/src/arrow/array/dictionary/value_map.rs

@@ -108,11 +108,12 @@ impl<K: DictionaryKey, M: MutableArray> ValueMap<K, M> {
             // safety: we only iterate within bounds
             let value = unsafe { values.value_unchecked_at(index) };
             let hash = ahash_hash(value.borrow());
-            match map.raw_entry_mut().from_hash(hash, |item| {
+            let entry = map.raw_entry_mut().from_hash(hash, |item| {
                 // safety: invariant of the struct, it's always in bounds since we maintain it
                 let stored_value = unsafe { values.value_unchecked_at(item.key.as_usize()) };
                 stored_value.borrow() == value.borrow()
-            }) {
+            });
+            match entry {
                 RawEntryMut::Occupied(_) => {
                     return Err(Error::InvalidArgumentError(
                         "duplicate value in dictionary values array".into(),
@@ -158,25 +159,25 @@ impl<K: DictionaryKey, M: MutableArray> ValueMap<K, M> {
         M::Type: Eq + Hash,
     {
         let hash = ahash_hash(value.as_indexed());
-        Ok(
-            match self.map.raw_entry_mut().from_hash(hash, |item| {
-                // safety: we've already checked (the inverse) when we pushed it, so it should be ok?
-                let index = unsafe { item.key.as_usize() };
-                // safety: invariant of the struct, it's always in bounds since we maintain it
-                let stored_value = unsafe { self.values.value_unchecked_at(index) };
-                stored_value.borrow() == value.as_indexed()
-            }) {
-                RawEntryMut::Occupied(entry) => entry.key().key,
-                RawEntryMut::Vacant(entry) => {
-                    let index = self.values.len();
-                    let key = K::try_from(index).map_err(|_| Error::Overflow)?;
-                    entry.insert_hashed_nocheck(hash, Hashed { hash, key }, ()); // NB: don't use .insert() here!
-                    push(&mut self.values, value)?;
-                    debug_assert_eq!(self.values.len(), index + 1);
-                    key
-                }
-            },
-        )
+        let entry = self.map.raw_entry_mut().from_hash(hash, |item| {
+            // safety: we've already checked (the inverse) when we pushed it, so it should be ok?
+            let index = unsafe { item.key.as_usize() };
+            // safety: invariant of the struct, it's always in bounds since we maintain it
+            let stored_value = unsafe { self.values.value_unchecked_at(index) };
+            stored_value.borrow() == value.as_indexed()
+        });
+        let result = match entry {
+            RawEntryMut::Occupied(entry) => entry.key().key,
+            RawEntryMut::Vacant(entry) => {
+                let index = self.values.len();
+                let key = K::try_from(index).map_err(|_| Error::Overflow)?;
+                entry.insert_hashed_nocheck(hash, Hashed { hash, key }, ()); // NB: don't use .insert() here!
+                push(&mut self.values, value)?;
+                debug_assert_eq!(self.values.len(), index + 1);
+                key
+            }
+        };
+        Ok(result)
     }
 
     pub fn shrink_to_fit(&mut self) {

src/common/arrow/src/arrow/array/fixed_size_list/mod.rs

@@ -26,7 +26,6 @@ mod data;
 mod ffi;
 pub(super) mod fmt;
 mod iterator;
-pub use iterator::*;
 mod mutable;
 pub use mutable::*;
 

src/common/arrow/src/arrow/array/map/mod.rs

@@ -27,7 +27,6 @@ mod data;
 mod ffi;
 pub(super) mod fmt;
 mod iterator;
-pub use iterator::*;
 
 /// An array representing a (key, value), both of arbitrary logical types.
 #[derive(Clone)]

src/common/arrow/src/arrow/array/primitive/mod.rs

@@ -35,7 +35,6 @@ mod ffi;
 pub(super) mod fmt;
 mod from_natural;
 mod iterator;
-pub use iterator::*;
 mod mutable;
 pub use mutable::*;
 

src/common/arrow/src/arrow/buffer/mod.rs

@@ -22,6 +22,7 @@ use std::ops::Deref;
 
 use crate::arrow::ffi::InternalArrowArray;
 
+#[allow(dead_code)]
 pub(crate) enum BytesAllocator {
     InternalArrowArray(InternalArrowArray),
 

src/common/arrow/src/arrow/compute/sort/row/variable.rs

@@ -59,7 +59,7 @@ pub fn encoded_len(a: Option<&[u8]>) -> usize {
 pub fn encode<'a, I: Iterator<Item = Option<&'a [u8]>>>(out: &mut Rows, i: I, opts: SortOptions) {
     for (offset, maybe_val) in out.offsets.iter_mut().skip(1).zip(i) {
         match maybe_val {
-            Some(val) if val.is_empty() => {
+            Some([]) => {
                 out.buffer[*offset] = match opts.descending {
                     true => !EMPTY_SENTINEL,
                     false => EMPTY_SENTINEL,

src/common/arrow/src/arrow/io/parquet/read/deserialize/nested_utils.rs

@@ -26,7 +26,6 @@ use super::super::Pages;
 use super::utils::DecodedState;
 use super::utils::MaybeNext;
 use super::utils::PageState;
-pub use super::utils::Zip;
 use crate::arrow::array::Array;
 use crate::arrow::bitmap::MutableBitmap;
 use crate::arrow::error::Result;

src/common/arrow/src/arrow/io/parquet/read/deserialize/primitive/basic.rs

@@ -321,7 +321,6 @@ impl<T, I, P, F> Iter<T, I, P, F>
 where
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {

src/common/arrow/src/arrow/io/parquet/read/deserialize/primitive/dictionary.rs

@@ -74,7 +74,6 @@ where
     K: DictionaryKey,
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {
@@ -153,7 +152,6 @@ where
     K: DictionaryKey,
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {

src/common/arrow/src/arrow/io/parquet/read/deserialize/primitive/integer.rs

@@ -226,7 +226,6 @@ impl<T, I, P, F> IntegerIter<T, I, P, F>
 where
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {

src/common/arrow/src/arrow/io/parquet/read/deserialize/primitive/nested.rs

@@ -187,7 +187,6 @@ pub struct NestedIter<T, I, P, F>
 where
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {
@@ -205,7 +204,6 @@ impl<T, I, P, F> NestedIter<T, I, P, F>
 where
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {
@@ -234,7 +232,6 @@ impl<T, I, P, F> Iterator for NestedIter<T, I, P, F>
 where
     I: Pages,
     T: NativeType,
-
     P: ParquetNativeType,
     F: Copy + Fn(P) -> T,
 {

src/common/arrow/src/arrow/types/simd/packed.rs

@@ -26,14 +26,14 @@ pub use std::simd::i8x8;
 pub use std::simd::mask32x16 as m32x16;
 pub use std::simd::mask64x8 as m64x8;
 pub use std::simd::mask8x64 as m8x64;
+pub use std::simd::prelude::SimdPartialEq;
 pub use std::simd::u16x32;
 pub use std::simd::u16x8;
 pub use std::simd::u32x16;
 pub use std::simd::u32x8;
 pub use std::simd::u64x8;
 pub use std::simd::u8x64;
 pub use std::simd::u8x8;
-pub use std::simd::SimdPartialEq;
 
 /// Vector of 32 16-bit masks
 #[allow(non_camel_case_types)]

src/common/arrow/src/lib.rs

@@ -13,6 +13,7 @@
 // limitations under the License.
 
 #![feature(iter_advance_by)]
+#![allow(clippy::unconditional_recursion)]
 #![cfg_attr(feature = "simd", feature(portable_simd))]
 #![allow(clippy::redundant_closure_call)]
 #![allow(clippy::non_canonical_partial_ord_impl)]

src/common/arrow/tests/it/arrow/array/binary/mutable.rs

@@ -53,7 +53,7 @@ fn from_iter() {
 
 #[test]
 fn from_trusted_len_iter() {
-    let data = vec![vec![0; 0], vec![1; 1], vec![2; 2]];
+    let data = [vec![0; 0], vec![1; 1], vec![2; 2]];
     let a: MutableBinaryArray<i32> = data.iter().cloned().map(Some).collect();
     assert_eq!(a.values().deref(), &[1u8, 2, 2]);
     assert_eq!(a.offsets().as_slice(), &[0, 0, 1, 3]);

src/common/base/src/runtime/runtime_tracker.rs

@@ -49,6 +49,7 @@ use std::fmt::Formatter;
 use std::future::Future;
 use std::mem::take;
 use std::pin::Pin;
+use std::ptr::addr_of_mut;
 use std::sync::atomic::AtomicI64;
 use std::sync::atomic::Ordering;
 use std::sync::Arc;
@@ -67,7 +68,7 @@ pub static GLOBAL_MEM_STAT: MemStat = MemStat::global();
 // For implemented and needs to call drop, we cannot use the attribute tag thread local.
 // https://play.rust-lang.org/?version=nightly&mode=debug&edition=2021&gist=ea33533387d401e86423df1a764b5609
 thread_local! {
-    static TRACKER: RefCell<ThreadTracker> = RefCell::new(ThreadTracker::empty());
+    static TRACKER: RefCell<ThreadTracker> = const { RefCell::new(ThreadTracker::empty()) };
 }
 
 #[thread_local]
@@ -231,7 +232,7 @@ impl ThreadTracker {
     /// `size` is the positive number of allocated bytes.
     #[inline]
     pub fn alloc(size: i64) -> Result<(), AllocError> {
-        let state_buffer = unsafe { &mut STAT_BUFFER };
+        let state_buffer = unsafe { &mut *addr_of_mut!(STAT_BUFFER) };
 
         // Rust will alloc or dealloc memory after the thread local is destroyed when we using thread_local macro.
         // This is the boundary of thread exit. It may be dangerous to throw mistakes here.
@@ -265,7 +266,7 @@ impl ThreadTracker {
     /// `size` is positive number of bytes of the memory to deallocate.
     #[inline]
     pub fn dealloc(size: i64) {
-        let state_buffer = unsafe { &mut STAT_BUFFER };
+        let state_buffer = unsafe { &mut *addr_of_mut!(STAT_BUFFER) };
 
         // Rust will alloc or dealloc memory after the thread local is destroyed when we using thread_local macro.
         if state_buffer.destroyed_thread_local_macro {

src/common/hashtable/src/lib.rs

@@ -12,6 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#![allow(internal_features)]
+#![allow(clippy::ptr_arg)]
 #![feature(core_intrinsics)]
 #![feature(allocator_api)]
 #![feature(arbitrary_self_types)]

src/common/io/src/bitmap.rs

@@ -17,7 +17,7 @@ use databend_common_exception::Result;
 use roaring::RoaringTreemap;
 
 pub fn parse_bitmap(buf: &[u8]) -> Result<RoaringTreemap> {
-    match std::str::from_utf8(buf)
+    std::str::from_utf8(buf)
         .map_err(|e| e.to_string())
         .and_then(|s| {
             let s: String = s.chars().filter(|c| !c.is_whitespace()).collect();
@@ -26,14 +26,17 @@ pub fn parse_bitmap(buf: &[u8]) -> Result<RoaringTreemap> {
                 .map(|v| v.parse::<u64>().map_err(|e| e.to_string()))
                 .collect();
             result
-        }) {
-        Ok(v) => {
-            let rb = RoaringTreemap::from_iter(v.iter());
-            Ok(rb)
-        }
-        Err(_) => Err(ErrorCode::BadBytes(format!(
-            "Invalid Bitmap value: {:?}",
-            String::from_utf8_lossy(buf)
-        ))),
-    }
+        })
+        .map_or_else(
+            |_| {
+                Err(ErrorCode::BadBytes(format!(
+                    "Invalid Bitmap value: {:?}",
+                    String::from_utf8_lossy(buf)
+                )))
+            },
+            |v| {
+                let rb = RoaringTreemap::from_iter(v.iter());
+                Ok(rb)
+            },
+        )
 }