Re: [boost] Interest in B-tree library for Boost?

Subject: Re: [boost] Interest in B-tree library for Boost?
From: Beman Dawes (bdawes_at_[hidden])
Date: 2010-09-15 15:20:13

On Wed, Sep 15, 2010 at 2:19 PM, Chad Nelson
<chad.thecomfychair_at_[hidden]> wrote:
> On Wed, 15 Sep 2010 09:44:42 -0700
> "Simonson, Lucanus J" <lucanus.j.simonson_at_[hidden]> wrote:
>
>> [...] Now that we have broken free of the 32bit addressable limit for
>> commodity hardware there is less need for self paging data structures
>> than there was a few years ago.  Unless there is a way to configure it
>> that gives a performance advantage over std::map we might have to wait
>> until 64 bits  (or 48 as the case may be) becomes too confining. [...]
>
> Not necessarily. There's quite a lot of need for an automatically
> *persistent* data structure like this -- i.e. one that survives between
> different runs of a program -- right now. Full-blown databases are
> overkill for many more-modest purposes, especially SQL databases.
>
> I realize that this could be done with serialization and such, and that
> would make sense if you needed to use the data on different
> architectures or operating systems...

The proposed B-tree library supplies traits classes for endian
implementation types, so that the data is portable between
architectures or operating systems if desired.  The needs of the
B-tree library were what motivated my work on the proposed Endian
library. That's already part of the B-tree code, and is passing the
test suite. I've also been hand inspecting dumps of disk data just to
be sure.

Cheers,

--Beman