mirror of
https://github.com/status-im/nim-stew.git
synced 2025-02-10 19:16:37 +00:00
638e7acc8e
Without lifetime tracking, ranges pose a number of issues with regards to memory safety - use openArray instead which is limited but safe. In its present form, ranges take a copy of whatever is passed to them which obscures the copies that they take, leading to misleading and slow code. If openArray is not applicable, simply do a seq copy - it's fine and really doesn't matter in 95% of all cases, then profile and use other techniques where needed.
35 lines
1.0 KiB
Nim
35 lines
1.0 KiB
Nim
{.deprecated: "unattractive memory unsafety - use openArray and other techniques instead".}
|
|
|
|
import
|
|
ptr_arith
|
|
|
|
type
|
|
MemRange* = object
|
|
start: pointer
|
|
size: csize
|
|
|
|
template len*(mr: MemRange): int = mr.size
|
|
template `[]`*(mr: MemRange, idx: int): byte = (cast[ptr byte](shift(mr.start, idx)))[]
|
|
proc baseAddr*(mr: MemRange): pointer = mr.start
|
|
|
|
proc makeMemRange*(start: pointer, size: csize): MemRange =
|
|
result.start = start
|
|
result.size = size
|
|
|
|
proc toMemRange*(x: string): MemRange =
|
|
result.start = x.cstring.pointer
|
|
result.size = x.len
|
|
|
|
proc toMemRange*[T](x: openarray[T], fromIdx, toIdx: int): MemRange =
|
|
doAssert(fromIdx >= 0 and toIdx >= fromIdx and fromIdx < x.len and toIdx < x.len)
|
|
result.start = unsafeAddr x[fromIdx]
|
|
result.size = (toIdx - fromIdx + 1) * T.sizeof
|
|
|
|
proc toMemRange*[T](x: openarray[T], fromIdx: int): MemRange {.inline.} =
|
|
toMemRange(x, fromIdx, x.high)
|
|
|
|
proc toMemRange*[T](x: openarray[T]): MemRange {.inline.} =
|
|
toMemRange(x, 0, x.high)
|
|
|
|
template toMemRange*(mr: MemRange): MemRange = mr
|