Raw data (was Re: Lexique full?)

Richard Durbin rd at sanger.ac.uk
Wed May 21 04:16:09 EST 2003


No, acedb always uses files, not raw data partitions.  We did ask about 
this at one point, and were told there would be little or no benefit in 
writing the additional code to use raw partitions.  Modern OS are pretty 
good with file I/O.

Richard

Steve Gaarder wrote:
> On 20 May 2003, Richard Durbin wrote:
> 
> 
>>Ah, sorry, this is really a hard limit in the current ACEDB code.  We
>>use 32 bits for a key for an object, 8 bits for the class and 24 for the
>>index of an object in the class.  It would be a very major restructuring
>>to change this.  You'll have to rethink your design I am afraid, or
>>rewrite the acedb kernel from the ground up.
>>
> 
> THanks,  I think redoing the data model will be the easier choice :-)
> 
> Another question: is support for raw data partitions (rather than files)
> implemented?  Does it hae any advantages for large data sets?
> 
> thanks,
> 
> Steve Gaarder
> 
> 





More information about the Acedb mailing list