Re: [rfc] Near-constant time directory index for Ext2

[Davide Libenzi <davidel@xmailserver.org>]

On 21-Feb-2001 Martin Mares wrote:
> Hello!
> 
>> My personal preference goes to skiplist coz it doesn't have fixed ( or
>> growing
>> ) tables to handle. You've simply a stub of data togheter with FS data in
>> each
>> direntry.
> 
> Another problem with skip lists is that they require variable sized nodes,
> so you either need to keep free chunk lists and lose some space in deleted
> nodes kept in these lists, or you choose to shift remaining nodes which is
> slow and complicated as you need to keep the inter-node links right. With
> hashing, you can separate the control part of the structure and the actual
> data and shift data while leaving most of the control part intact.

An entry in skip list table is a u32 direntry offset and You've not to keep
free entries, simply the height of the node will change depending on the number
of entries.


>> And performance ( O(log2(n)) ) are the same for whatever number of entries.
> 
> I don't understand this complexity estimate -- it cannot be the same for
> whatever number of entries as the complexity function depends on the number
> of entries.

n == number of entries

For constant I mean the formula not the result.



- Davide