Thread

1. Re: [PATCH] ANALYZE: hash-accelerate MCV tracking for equality-only types

   Ilia Evdokimov <ilya.evdokimov@tantorlabs.com> — 2026-05-03T14:16:24Z

   On 4/14/26 04:34, Chengpeng Yan wrote:
   
   > I split v5 accordingly. The first patch changes the singleton handling
   > from shifting to a cursor-based eviction scheme, and the second patch
   > adds the hash lookup.
   I reviewed v5 of the patches. Instead of going through each issue one by 
   one, I made a pass to clean up and clarify the code and summarize the 
   main changes belows:
   
   - Fixed a few typos in comments and added comments in places where the 
   logic was not immediately clear;
   
   - Rewrote the bubble-up loop using `for` loop, which I find more 
   readable. Also removed some confusing uses of the `j` variable that mase 
   the flow harder to follow;
   
   - Simplified parts of the code to improve overall readability.
   
   -- 
   Best regards,
   Ilia Evdokimov,
   Tantor Labs LLC,
   https://tantorlabs.com/
   

2. Re: [PATCH] ANALYZE: hash-accelerate MCV tracking for equality-only types

   John Naylor <johncnaylorls@gmail.com> — 2026-05-04T02:10:16Z

   On Thu, Feb 26, 2026 at 6:46 PM Ilia Evdokimov
   <ilya.evdokimov@tantorlabs.com> wrote:
   > While reviewing the patch more closely, I noticed
   > that compute_distinct_stats() is only used for types where we have =, !=
   > but not <. In practice, most common scalar types go through
   > compute_scalar_stats() instead.
   >
   > That makes me wonder how often this optimization would actually trigger
   > in real workloads. Since compute_scalar_stats() is the more common path,
   > there's chance that the hash-table based improvement in
   > compute_distinct_stats() may not provide a noticeable overall benefit.
   
   Coming back to this point, in any installation, the common path is
   going to vastly outnumber the rare path, so this patch is optimizing
   the wrong thing.
   
   -- 
   John Naylor
   Amazon Web Services
   
   
   
   

3. Re: [PATCH] ANALYZE: hash-accelerate MCV tracking for equality-only types

   Chengpeng Yan <chengpeng_yan@outlook.com> — 2026-05-06T04:51:45Z

   Hi,
   
   > On May 3, 2026, at 22:16, Ilia Evdokimov <ilya.evdokimov@tantorlabs.com> wrote:
   > 
   > I reviewed v5 of the patches. Instead of going through each issue one by one, I made a pass to clean up and clarify the code and summarize the main changes belows:
   > - Fixed a few typos in comments and added comments in places where the logic was not immediately clear;
   > - Rewrote the bubble-up loop using `for` loop, which I find more readable. Also removed some confusing uses of the `j` variable that mase the flow harder to follow;
   > - Simplified parts of the code to improve overall readability.
   
   Thanks for the review and for working on this.
   
   After giving it more thought, I agree with John’s point that the set of
   types benefiting from this optimization is currently too limited. While
   this optimization can still provide noticeable improvements in some
   specific cases, those cases do not seem common enough at this stage, so
   the overall benefit also seems fairly small.
   
   Because of that, I marked this patch as Withdrawn in the current
   CommitFest and plan to pause further work on it for now.
   
   That said, if postgres or some extensions introduce more commonly used
   types that are a good fit for this optimization, I think it would make
   sense to revisit and potentially revive this work.
   
   Thanks again for the effort and contribution.
   
   --
   Best regards,
   Chengpeng Yan