builtins: deflake TestSerialNormalizationWithUniqueUnorderedID #91868
Conversation
|
This change is |
|
|
||
| tdb.QueryRow(t, ` | ||
| WITH boundaries AS ( | ||
| SELECT i << (60) AS p FROM ROWS FROM (generate_series(0, 7)) AS t (i) |
There was a problem hiding this comment.
at a high level i get that this is computing a histogram, but want to add some comments about what's up with the bit-shifting and why the numbers 60, 62, and 7 were chosen? perhaps explain the SELECT p AS low, lead(p) OVER () AS high FROM boundaries window clause a bit too
| @@ -121,16 +121,30 @@ CREATE TABLE t ( | |||
| // Enforce 3 bits worth of range splits in the high order to collect range | |||
There was a problem hiding this comment.
does the comment about enforcing range splits still technically apply?
The use of range stats and splits was not very stable and not needed. Fixes cockroachdb#91858 Release note: None
ajwerner
force-pushed
the
ajwerner/fix-TestSerialNormalizationWithUniqueUnorderedID
branch
from
8a7541c to
fd1707a
Compare
There was a problem hiding this comment.
i think it needs a rebase and unskip for #91857
Done.
Reviewable status:
complete! 0 of 0 LGTMs obtained (waiting on @rafiss)
pkg/sql/sem/builtins/builtins_test.go line 121 at r1 (raw file):
Previously, rafiss (Rafi Shamim) wrote…
does the comment about enforcing range splits still technically apply?
Done.
pkg/sql/sem/builtins/builtins_test.go line 132 at r1 (raw file):
Previously, rafiss (Rafi Shamim) wrote…
at a high level i get that this is computing a histogram, but want to add some comments about what's up with the bit-shifting and why the numbers
60,62, and7were chosen? perhaps explain theSELECT p AS low, lead(p) OVER () AS high FROM boundarieswindow clause a bit too
Done. I generalized the query a bit to take a bits parameter and tried to explain it better.
|
bors r+ |
|
Build succeeded: |
The use of range stats and splits was not very stable and not needed.
Fixes #91858
Release note: None