> On Sep 24, 2024, at 10:50 AM, Peter Eisentraut <pe...@eisentraut.org> wrote:
> 
> Next, I have reviewed patches
> 
> v17-0010-Track-sort-direction-in-SortGroupClause.patch
> v17-0011-Track-scan-reversals-in-MergeJoin.patch
> 
> Both of these seem ok and sensible to me.
> 
> They take the concept of the "reverse" flag that already exists in the 
> affected code and just apply it more consistently throughout the various code 
> layers, instead of relying on strategy numbers as intermediate storage.  This 
> is both helpful for your ultimate goal in this patch series, and it also 
> makes the affected code areas simpler and more consistent and robust.
> 

Thanks for the review!

Yes, I found the existing use of a btree strategy number rather than a boolean 
"reverse" flag made using the code from other index AMs needlessly harder.  I 
am glad you see it the same way.

—
Mark Dilger
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company





Reply via email to