Trait nalgebra::base::storage::RawStorage [−][src]
pub unsafe trait RawStorage<T, R: Dim, C: Dim = U1>: Sized {
type RStride: Dim;
type CStride: Dim;
fn ptr(&self) -> *const T;
fn shape(&self) -> (R, C);
fn strides(&self) -> (Self::RStride, Self::CStride);
fn is_contiguous(&self) -> bool;
unsafe fn as_slice_unchecked(&self) -> &[T]ⓘ;
fn linear_index(&self, irow: usize, icol: usize) -> usize { ... }
fn get_address_unchecked_linear(&self, i: usize) -> *const T { ... }
fn get_address_unchecked(&self, irow: usize, icol: usize) -> *const T { ... }
unsafe fn get_unchecked_linear(&self, i: usize) -> &T { ... }
unsafe fn get_unchecked(&self, irow: usize, icol: usize) -> &T { ... }
}
Expand description
The trait shared by all matrix data storage.
TODO: doc
In generic code, it is recommended use the Storage
trait bound instead. The RawStorage
trait bound is generally used by code that needs to work with storages that contains
MaybeUninit<T>
elements.
Note that Self
must always have a number of elements compatible with the matrix length (given
by R
and C
if they are known at compile-time). For example, implementors of this trait
should not allow the user to modify the size of the underlying buffer with safe methods
(for example the VecStorage::data_mut
method is unsafe because the user could change the
vector’s size so that it no longer contains enough elements: this will lead to UB.
Associated Types
Required methods
The dimension of the matrix at run-time. Arr length of zero indicates the additive identity
element of any dimension. Must be equal to Self::dimension()
if it is not None
.
The spacing between consecutive row elements and consecutive column elements.
For example this returns (1, 5)
for a row-major matrix with 5 columns.
fn is_contiguous(&self) -> bool
fn is_contiguous(&self) -> bool
Indicates whether this data buffer stores its elements contiguously.
Safety
This function must not return true
if the underlying storage is not contiguous,
or undefined behaviour will occur.
Retrieves the data buffer as a contiguous slice.
Safety
The matrix components may not be stored in a contiguous way, depending on the strides. This method is unsafe because this can yield to invalid aliasing when called on some pairs of matrix slices originating from the same matrix with strides.
Call the safe alternative matrix.as_slice()
instead.
Provided methods
fn linear_index(&self, irow: usize, icol: usize) -> usize
fn linear_index(&self, irow: usize, icol: usize) -> usize
Compute the index corresponding to the irow-th row and icol-th column of this matrix. The index must be such that the following holds:
let lindex = self.linear_index(irow, icol);
assert!(*self.get_unchecked(irow, icol) == *self.get_unchecked_linear(lindex))
fn get_address_unchecked_linear(&self, i: usize) -> *const T
fn get_address_unchecked_linear(&self, i: usize) -> *const T
Gets the address of the i-th matrix component without performing bound-checking.
Safety
If the index is out of bounds, dereferencing the result will cause undefined behavior.
fn get_address_unchecked(&self, irow: usize, icol: usize) -> *const T
fn get_address_unchecked(&self, irow: usize, icol: usize) -> *const T
Gets the address of the i-th matrix component without performing bound-checking.
Safety
If the index is out of bounds, dereferencing the result will cause undefined behavior.
unsafe fn get_unchecked_linear(&self, i: usize) -> &T
unsafe fn get_unchecked_linear(&self, i: usize) -> &T
Retrieves a reference to the i-th element without bound-checking.
Safety
If the index is out of bounds, the method will cause undefined behavior.