> On Tue, Feb 07, 2023 at 11:14:52PM +0300, Sergei Kornilov wrote:
> Hello!
Thanks for reviewing.
> Unfortunately, rebase is needed again due to recent changes in queryjumblefuncs (
9ba37b2cb6a174b37fc51d0649ef73e56eae27fc)
Yep, my favourite game, rebaseball. Will post a new version soon, after
figuring out all the recent questions.
> It seems a little strange to me that with const_merge_threshold = 1, such a test case gives the same result as with
const_merge_threshold= 2
>
> select pg_stat_statements_reset();
> set const_merge_threshold to 1;
> select * from test where i in (1,2,3);
> select * from test where i in (1,2);
> select * from test where i in (1);
> select query, calls from pg_stat_statements order by query;
>
> query | calls
> -------------------------------------+-------
> select * from test where i in (...) | 2
> select * from test where i in ($1) | 1
>
> Probably const_merge_threshold = 1 should produce only "i in (...)"?
Well, it's not intentional, probably I need to be more careful with
off-by-one. Although I agree to a certain extent with Peter questioning
the value of having numerical option here, let me think about this.
> const_merge_threshold is "the minimal length of an array" (more or equal) or "array .. length is larger than" (not
equals)?I think the documentation is ambiguous in this regard.
>
> I also noticed a typo in guc_tables.c: "Sets the minimal numer of constants in an array" -> number
Yep, I'll rephrase the documentation.