On Wednesday, 23 July 2014 at 00:59:38 UTC, H. S. Teoh via
Digitalmars-d wrote:
I think it's best to keep a bit count of the number of valid
bits at the
end of the array, in addition to the number of words in the
underlying
array. You might also want to consider keeping a bit count of
the num
On Wednesday, 23 July 2014 at 02:33:39 UTC, safety0ff wrote:
In brief, you're suggesting to always realloc on extension, no
matter if it's a sub-word extension.
This solves the stomping issue nicely
Actually, you'd need to allocate and copy on each extension, not
realloc, because realloc wo
On Wednesday, 23 July 2014 at 02:33:39 UTC, safety0ff wrote:
but it would cause a lot of GC churn in a concatenation loop.
Nevermind this. :o)
On Wednesday, 23 July 2014 at 00:59:38 UTC, H. S. Teoh via
Digitalmars-d wrote:
You might want to consider implementing a way of tracking how
many bits
in the final word are valid. That way, you can correctly trigger
reallocation if the user tries to write to a bit beyond the
current end
of t
On Tue, Jul 22, 2014 at 05:57:48PM -0700, H. S. Teoh via Digitalmars-d wrote:
[]
> You might want to consider implementing a way of tracking how many
> bits in the final word are valid. That way, you can correctly trigger
> reallocation if the user tries to write to a bit beyond the current
> e
On Tue, Jul 22, 2014 at 09:16:35PM +, safety0ff via Digitalmars-d wrote:
> Currently the implementation of BitArray's length setter is badly broken.
> Its implementation contains this code segment:
>
> if (newdim != olddim) {
> auto b = ptr[0 .. olddim];
> b.length = newdim
Currently the implementation of BitArray's length setter is badly
broken.
Its implementation contains this code segment:
if (newdim != olddim) {
auto b = ptr[0 .. olddim];
b.length = newdim;// realloc
ptr = b.ptr;
if (newdim & (bitsPerSizeT-1)) // Set any
