Re: [boost] Re: B-tree for handling hugh datastructures (partly on disk)

From: Jeremy Maitin-Shepard (jbms_at_[hidden])
Date: 2003-12-15 20:27:35

David Abrahams <dave_at_[hidden]> writes:

> Jeremy Maitin-Shepard <jbms_at_[hidden]> writes:
>> "Toon Knapen" <toon.knapen_at_[hidden]> writes:
>> 
>>> A few months ago there was some initial discussion about implementing a
>>> B-tree to handle hugh datastructures by storing parts of the B-tree on
>>> disk. 
>> 
>>> Could someone tell me if there is still something going in this area
>>> ? 
>> 
>> Such a library would probably depend on a serialization library
>> (i.e. Robert Ramey's).

> Wow, I really hope not.  Why make a low-level component like a B-tree
> dependent on serialization?  I can see why it'd be important for them
> to interoperate well, but why dependency?

There needs to be some mechanism to store and retrieve keys (and values)
to and from the disk.  If Robert's serialization library is not used,
some other library would need to be used.  I admit that a number of
limitations would have to be dealt with though, since the size of each
serialized key would need to be the same (or be bounded), and then a
value for M (where the resultant B-tree is M-ary) and L (where there are
L keys stored per node/disk block) would need to be calculated based on
the attributes of the storage device.

>> On a related note, I noticed that the Yahoo! boost files section is
>> ``members''-only, so perhaps someone could add my Yahoo! user to that
>> group (jmaitinshepard).

> You can sign up for the group yourself, AFAIK.

Ah, okay.

-- 
Jeremy Maitin-Shepard