BitArray/BitFields - Review

[Dmitry Olshansky]

On 29-Jul-12 01:20, Era Scarecrow wrote:
> On Saturday, 28 July 2012 at 20:59:15 UTC, Dmitry Olshansky wrote:
>> Great! But I strongly suggest to repost it in d.D newsgroup as it has
>> wider audience and is more appropriate for reviews.
>
>   I was thinking of not bothering Andrei or Walter while i fixed other
> issues on it before bringing more major ones forward (which is why i
> brought it to D.learn), but re-posting it there is easy enough
>
>> My solution is make slices different type e.g. BitSlice. It's always
>> reference to the original BitArray. Array itself still can be either
>> compact or not. All operations with slice go to array (and still check
>> if they can use word-aligned speed up).
>>
>> How does it sound?
>
>   Someone else also suggested making the slice different (Meaning a new
> Bitarray always is at max compact size).

Obviously that was me (blackwhale) ;)

Since we need at least a single
> flag to determine if the array is compact or array reference, there
> would be a larger loss which the current offset/maxOffset handle quite
> nicely filling in that space.

Not sure I followed you - what with offset/maxOffset ? You mean less 
space available for compact version?

  Making BitSlice separate suggests BitArray
> would be the bare minimum (canUseBulk, length, and index operations)
> while bitSlice would handle all others. But if you do that, either
> BitSlice would have to be inherently convertible to BitArray if you want
> to pass it to a function that expects a BitArray.

Take a look at std.container from what I see there Array does have a 
separate range type.

I'm hesitant to do it,
> but if there's a strong enough argument for it I wouldn't refuse it.
>
>> opCat isn't it just operator "~" and "~=" ? You can use opOpAssign for
>> "~=" IRC.
>
>   Yes, I think opAssign is used for ~= already, but doing the following
> requires opCat and opCat_r.
>
>   BitArray a; //filled with something
>   auto x = true ~ a;
>   auto y = a ~ true;
>
>   I just can't get it to accept it (for whatever reason). Compiler bug?
>
Hm opBinaryRight with if (op == "~") should work.

>> I suspect to!string wasn't CTFEable. Regardless you can catch a word
>> from Andrei Alexandrescu on the main D newsgroup who (I think) came up
>> with it in the first place.
>
>   If we can use the std.string functions I'll replace them all :) (As I
> was just following the theme since they were present).
>
>   I'll move this to the D group and see what we get for this.


-- 
Dmitry Olshansky