vibe.d + dpq2 + json trick

[Serg Gini]

On Wednesday, 20 August 2025 at 16:06:10 UTC, Robert Schadek 
wrote:
> On Wednesday, 20 August 2025 at 14:10:31 UTC, Denis Feklushkin 
> wrote:
>>
>> Also, I added PREPARE, row-by-row read for /normal (just for 
>> check if it helps) and replaced "id" by [0] on column 
>> accessing code
>>
>> Nothing really helped
>>
>> My measurements:
>>
>> ab -n 10000 -c 10 localhost:8080/oldschool Requests per 
>> second:    675.46 [#/sec] (mean)
>> ab -n 10000 -c 10 localhost:8080/normal Requests per second:   
>>  163.55 [#/sec] (mean)
>> ab -n 10000 -c 10 localhost:8080/special Requests per second:  
>>   3636.42 [#/sec] (mean)
>
> I haven't checked by I bet postgres caches parsed/optimized 
> queries. And the test would have a nearly perfect cache hit 
> ratio

I wonder with values will valkey show.
Another thing that ab tool will send the same request and the 
same ID will be requested over and over again.

Maybe if create let's say 1000 ids and every request then 
calculating random(1,1000) and use it for data extraction, it 
will show a bit more realistic values.