List:Internals« Previous MessageNext Message »
From:Michael Widenius Date:August 5 2008 10:41pm
Subject:Re: Dynamic record sizes for HEAP engine
View as plain text  

>>>>> "Harrison" == Harrison Fisk <harrison@stripped> writes:

Harrison> Hi,
Harrison> On Aug 2, 2008, at 2:18 PM, Michael Widenius wrote:

>> Hi!
>>>>>>> "Igor" == Igor Chernyshev <igor_cc75@stripped>
> writes:
Igor> Update for those interested in this patch --
Igor> Larry Zhou from Google and I have made a few bug fixes.
Igor> The diff and zip files have been updated.
Igor> Original diff can be found in "Deprecated" downloads.
Igor> It's still based on MySQL 5.0.45.
>> Thanks!
>> I had an discussion about this patch with Brian two days ago. He is
>> porting the patch to drizzle (basicly MySQL 5.1) and when he is ready
>> I will get a copy of it and plan to add it to the MySQL-5.1-maria
>> tree.

Harrison> I finished this yesterday and it is now part of the drizzle tree.  It  
Harrison> was relatively straight forward, just had a few bigger conflicts with  
Harrison> some of the 5.1 changes for reducing contention around internal  
Harrison> temporary tables.

Harrison> If I get some spare time, I hope to add the ability to not require the  
Harrison> key columns to be first in the table.  Once that is done, I wanted to  
Harrison> try to enable it for internal temporary tables (except those with  
Harrison> blobs) and do some benchmarks to see how it does compared to the fixed  
Harrison> width allocation.

Harrison> Those are more of long term plans since I am only doing this in my own  
Harrison> time.  Dunno when/if they will be done.

Hm; The restriction of having key columns first in a table is
something that must be lifted before I can put it into 5.1-maria to
replace heap tables for normal users.

Will look into this as soon as I have some time to spear.

Re: Dynamic record sizes for HEAP engineIgor Chernyshev31 Jul
  • Re: Dynamic record sizes for HEAP engineMichael Widenius2 Aug
    • Re: Dynamic record sizes for HEAP engineHarrison Fisk5 Aug
      • Re: Dynamic record sizes for HEAP engineMichael Widenius6 Aug