# Changelog
## 1.2.2
* **Soundness:** Previous versions of the iterators in
this crate (since 0.4) had a math error in the `nth`
method, causing them to potentially go out of bounds.
## 1.2
* The `Safe` and `Unsafe` types now also derive `Default`, `Clone`, and `Copy`.
This doesn't do too much since they're already ZSTs with a public constructor,
but it doesn't hurt.
* `VolAddress`: Added const fn `as_ptr` and `as_mut_ptr`.
* `VolBlock`: Added const fn `as_usize`, `as_ptr`, `as_mut_ptr`, and non-const
fn `as_slice_ptr` and `as_slice_mut_ptr`.
* It turns out that getting the pointer to a `VolBlock` by indexing to the 0th
element and then turning that into a usize and then turning that into a
pointer was enough layers to confuse LLVM. Specifically, all volatile accesses
have an "aligned and non-null" debug check, which wasn't getting optimized out
of debug builds with `build-std`, even with `opt-level=3`. Providing these
more direct conversion methods does seem to help LLVM eliminate that non-null
check more often.
* Added `core::fmt::Pointer` impls. While `Debug` formats the address along with
extra metadata, `Pointer` just formats the address.
## 1.1
* Added `VolAddress::as_volblock` for (unsafely) converting from a `VolAddress`
to an array into a `VolBlock`. This is totally fine in any case I've ever
seen, but the general policy of the crate is that any creation of a `VolBlock`
be an unsafe action, and so this is unsafe for consistency.
* Also adds the `experimental_volmatrix` cargo feature, which adds another
opt-in type for people to experiment with.
## 1.0.2
* Temporarily adds the `experimental_volregion` feature, allowing a person to
*experimentally* opt-in to the `VolRegion` type. This type will be part of a
1.1 release of the crate at some point. This feature is **not** part of the
crate's SemVer, and it will go away entirely once `VolRegion` becomes a stable
part of the crate.
## 1.0.0
* Initial stable release.