Repository navigation
lpad, rpad, translate should use codepoints, not graphemes #21060
Description
Describe the bug
For consistency with most other SQL implementations (including Postgres), as well as the rest of DataFusion, lpad, rpad, and translate should use codepoints when manipulating Unicode strings, not graphemes. #3054 switched to using codepoints instead of graphemes in most places, but AFAICS the same should be done for these remaining UDFs.
To Reproduce
No response
Expected behavior
No response
Additional context
No response
Activity
neilconway commentedon Mar 19, 2026
take
neilconway commentedon Mar 19, 2026
@Dandandan I realize it was a long time ago, but if there was a reason you chose to have
lpad,rpad, andtranslatecontinue to consider graphemes instead of codepoints, please let me know 😊fix: Use codepoints in `lpad`, `rpad`, `translate` (#21405)
fix: Use codepoints in `lpad`, `rpad`, `translate` (apache#21405)