diff --git a/src/error.rs b/src/error.rs
index 5b7b642..9cb2a90 100644
--- a/src/error.rs
+++ b/src/error.rs
@@ -52,6 +52,4 @@ impl Display for Error {
         f.write_str("bitcode error")
     }
 }
-#[cfg(feature = "std")]
-// TODO expose to no_std when error_in_core stabilized (https://github.com/rust-lang/rust/issues/103765)
-impl std::error::Error for Error {}
+impl core::error::Error for Error {}
diff --git a/src/lib.rs b/src/lib.rs
index d82e0ee..09640a8 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -26,7 +26,6 @@ mod fast;
 mod histogram;
 mod int;
 mod length;
-mod nightly;
 mod pack;
 mod pack_ints;
 mod pack_shared;
diff --git a/src/nightly.rs b/src/nightly.rs
deleted file mode 100644
index 2215297..0000000
--- a/src/nightly.rs
+++ /dev/null
@@ -1,18 +0,0 @@
-/// `#![feature(int_roundings)]` was stabilized in 1.73, but we want to avoid MSRV that high.
-macro_rules! impl_div_ceil {
-    ($name:ident, $t:ty) => {
-        #[inline(always)]
-        #[allow(unused_comparisons)] // < 0 checks not required for unsigned
-        pub const fn $name(lhs: $t, rhs: $t) -> $t {
-            let d = lhs / rhs;
-            let r = lhs % rhs;
-            if (r > 0 && rhs > 0) || (r < 0 && rhs < 0) {
-                d + 1
-            } else {
-                d
-            }
-        }
-    };
-}
-impl_div_ceil!(div_ceil_u8, u8);
-impl_div_ceil!(div_ceil_usize, usize);
diff --git a/src/pack.rs b/src/pack.rs
index 2a47746..bf79918 100644
--- a/src/pack.rs
+++ b/src/pack.rs
@@ -44,7 +44,7 @@ impl PackingTrait for Packing {
 impl Packing {
     fn read(input: &mut &[u8]) -> Result<(Self, bool)> {
         let v = consume_byte(input)?;
-        let p_u8 = crate::nightly::div_ceil_u8(v, 2);
+        let p_u8 = v.div_ceil(2);
         let offset_by_min = v & 1 != 0;
         let p = match p_u8 {
             0 => Self::_256,
@@ -280,7 +280,7 @@ pub fn unpack_bytes_less_than<'a, const N: usize, const HISTOGRAM: usize>(
             check_less_than_u8::<N, HISTOGRAM, FACTOR>(out)
         } else {
             let floor = length / divisor;
-            let ceil = crate::nightly::div_ceil_usize(length, divisor);
+            let ceil = length.div_ceil(divisor);
             let whole = &original_input[..floor];
 
             // Can only `partial_with_garbage % FACTOR` partial_length times as the rest are undefined garbage.
@@ -335,6 +335,7 @@ pub fn unpack_bytes_less_than<'a, const N: usize, const HISTOGRAM: usize>(
                 })
             };
             if let Ok(h) = histogram {
+                debug_assert!(check_less_than_u8::<N, 0, FACTOR>(out).is_ok());
                 debug_assert_eq!(
                     h,
                     check_histogram(crate::histogram::histogram(out)).unwrap()
@@ -439,7 +440,7 @@ fn unpack_arithmetic<const FACTOR: usize>(
 
     // TODO STRICT: check that packed.all(|&b| b < FACTOR.powi(divisor)).
     let floor = unpacked_len / divisor;
-    let ceil = crate::nightly::div_ceil_usize(unpacked_len, divisor);
+    let ceil = unpacked_len.div_ceil(divisor);
     let packed = consume_bytes(input, ceil)?;
 
     debug_assert!(out.is_empty());
@@ -666,15 +667,18 @@ mod tests {
     bench_n!(bench_unpack_arithmetic, 2, 3, 4, 6, 16);
 
     fn test_pack_bytes_less_than_n<const N: usize, const FACTOR: usize>() {
-        for n in [1, 11, 97, 991, 10007].into_iter().flat_map(|n_prime| {
-            let divisor = if FACTOR == 256 {
-                1
-            } else {
-                super::factor_to_divisor::<FACTOR>()
-            };
-            let n_factor = crate::nightly::div_ceil_usize(n_prime, divisor) * divisor;
-            [n_factor, n_prime]
-        }) {
+        for n in [1usize, 11, 97, 991, 10007]
+            .into_iter()
+            .flat_map(|n_prime| {
+                let divisor = if FACTOR == 256 {
+                    1
+                } else {
+                    super::factor_to_divisor::<FACTOR>()
+                };
+                let n_factor = n_prime.div_ceil(divisor) * divisor;
+                [n_factor, n_prime]
+            })
+        {
             let bytes: Vec<_> = crate::random_data(n)
                 .into_iter()
                 .map(|v: usize| (v % N as usize) as u8)
diff --git a/src/pack_ints.rs b/src/pack_ints.rs
index d9c528f..d73c503 100644
--- a/src/pack_ints.rs
+++ b/src/pack_ints.rs
@@ -43,7 +43,7 @@ impl PackingTrait for Packing {
 impl Packing {
     fn read<T: SizedUInt>(input: &mut &[u8]) -> Result<(Self, bool)> {
         let v = consume_byte(input)?;
-        let p_u8 = crate::nightly::div_ceil_u8(v, 2) + Self::new(T::MAX) as u8;
+        let p_u8 = v.div_ceil(2) + Self::new(T::MAX) as u8;
         let offset_by_min = v & 1 != 0;
         let p = match p_u8 {
             0 => Self::_128,
diff --git a/src/serde/mod.rs b/src/serde/mod.rs
index 0d49bce..a8356ee 100644
--- a/src/serde/mod.rs
+++ b/src/serde/mod.rs
@@ -46,9 +46,6 @@ fn get_mut_or_resize<T: Default>(vec: &mut Vec<T>, index: usize) -> &mut T {
     unsafe { vec.get_unchecked_mut(index) }
 }
 
-#[cfg(not(feature = "std"))]
-impl serde::ser::StdError for Error {}
-
 impl serde::ser::Error for Error {
     fn custom<T>(t: T) -> Self
     where