Vik Fearing
banner
vik.thegresqlpost.org
Vik Fearing
@vik.thegresqlpost.org
Member of the ISO/IEC working group for SQL. Performance is an implementation detail.

HUGE fan of MLB and WFTDA, and pretty much not interested in any other sport.

he/him by default but will happily answer to anything.

A cheeseburger in paradise. #IYKYK
I have no idea what that means. 😔
November 13, 2025 at 3:22 PM
Prefix shenanigans?
November 13, 2025 at 3:20 PM
There are many reasons to hate #SQL but also many reasons to love it. May I ask what prompted this post? Is it something that could be improved?
November 13, 2025 at 3:01 PM
I once put a sign "I am not here today" on my open plan desk, and put my earphones on. I kept getting interrupted and asked what I meant by not being there because I was obviously there.
November 13, 2025 at 1:52 PM
I don't see how. Consider:

CREATE TABLE t (col INTEGER ARRAY[0]);

Now I have a column that only accepts ARRAY[]. I can't put anything else in it.
November 12, 2025 at 11:56 AM
November 2, 2025 at 11:23 AM
If you're parsing postgres's text representation of arrays, are you handling custom separators?

For example, the box type uses ';' instead of ',' and any custom type can choose their own.
October 13, 2025 at 11:53 AM
Now that you know it exists, you have to implement it in @jooq.org. I don’t make the rules.
October 11, 2025 at 12:18 PM
The bounds get lost during pretty much any operation. One day I will go spelunking to see why this exists in the first place.

#PostgreSQLQuirks
October 11, 2025 at 10:18 AM
My indexing starts where I tell it to.

VALUES ((CAST('[-1:1]={1,2,3}' AS INTEGER ARRAY))[0])

=> 2
October 10, 2025 at 1:27 PM
Reposted by Vik Fearing
Also Claude, 🔥
October 8, 2025 at 8:54 AM
Yes. You can change the default storage properties for a lot of data types, but it only operates on single values.
October 7, 2025 at 7:33 AM
They are not compressed. Certain values might be (see TOAST) but not the page itself.
October 6, 2025 at 8:27 PM
Then what is it?
September 26, 2025 at 4:43 PM
Which is exactly the wrong time to start testing. At the very latest you should have been testing RC1.
September 26, 2025 at 4:01 PM
Why would you only start testing now? You should have been testing weeks ago!
September 26, 2025 at 5:57 AM