Re: [boost] [ boost ] [ Trie ]

Meant to ask if such a specialization had been added or was planned... On Fri, Apr 3, 2015 at 5:05 AM, Kenneth Adam Miller < kennethadammiller@gmail.com> wrote:

I know this might be a little bit late in asking, because I've been insanely busy too.

Say I know that I have a finite length of my tree-just 3 bytes-that I would like to be cached. Instead of resolving the tree dynamically, I would like there to be a contiguous array, but for each array entry, have some kind special value-either a mapping that produces a value despite length, or a concrete length limited pointer to result type.

[ <- 32 megabytes -> ]

An index into the radix tree can be performed by multiplying the first byte by 2^16 (possibly minus one). Then a pointer to some dynamically managed tree would suit our needs.

I calculate that a lookup, assuming that the tree is also shallow, shouldn't take more than just a handful of instructions. I'm all about speed in this scenario, and my use case requites that insertion consume as few instructions as feasibly possible. Other operations aren't as important.

On Thu, Apr 2, 2015 at 7:09 PM, Cosmin Boaca

...

wrote:

...

Hello,

I'm sorry but I had an insane coursework load during the last two weeks. I had about 4 programming assignments to solve based on system programming on both linux / windows and I hadn't any time to work at Boost.Trie. However, I have finished all my assignments and I will work on the iterator tomorrow and maybe during the weekend. I hope I will have it working until Sunday.

Cosmin

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost