maria-developers team mailing list archive

[Igor Babaev <igor@xxxxxxxxxxxx>]

Another consideration:

They use 'segments' when talk about different parts of something.
Webster's dictionary gives you an example: <all segments of the
population agree>.
Partitions are assumed to be always similar.

In the future new models of key cache will appear for sure, in
particular those ones where parts of the key cache will behave
themselves differently.

Regards,
Igor.

Henrik Ingo wrote:
> On Mon, May 17, 2010 at 9:14 PM, Igor Babaev <igor@xxxxxxxxxxxx> wrote:
>> 1. the author of the original patch
>> 2. our community members (not only PeterZ!)
>> 3. our developers
>>
>> what term they prefer 'Segmented Key Cache' or 'Partitioned Key Cache'.
>>
>> To be quite honest I also should mention that MichaelR from Oracle/MySQL
>> introduced in 5.5 a possibility for MySQL partitions in MyISAM to use
>> different key caches for different partitions. Before that all
>> partitions could use only default key caches. This feature is 100%
>> orthogonal to the discussed key cache partitioning.
> 
> This was the only reason for the rename. (It is also the reason the
> rename was done after the implementation, and very fast, because we
> had to decide on a name before Monty's keynote.)
> 
> So the main issue is to choose a name that will avoid confusion with
> the feature Mikael R has implemented. I think anything that achieves
> this can still be considered.
> 
> In any case, it is not the intent to change anything in the code,
> except I guess we need to syncronize the name of the feature with the
> user visible variable names.
> 
> henrik