Uh oh!
There was an error while loading. Please reload this page.
[Branch-50] Backport: Support Decimal32/64 types (#17501) - #17907
Conversation
alamb
commented
Oct 3, 2025
Why does this PR has substantially more lines that the original PR? This PR Original PR: #17501  |
AdamGS
commented
Oct 3, 2025
there are changes in |
* Support Decimal32/64 types * Fix bugs, tests, handle more aggregate functions and schema * Fill out more parts in expr,common and expr-common * Some stragglers and overlooked corners * Actually commit the avg_distinct support --------- Co-authored-by: Andrew Lamb <andrew@nerdnetworks.org>
3bb79ae to
d8b844dCompareAdamGS
commented
Oct 3, 2025
Ok now it seems better, its not a perfect match because some of the changes were based upon #17459, which wasn't backported. |
| /// 64bit float | ||
| Float64(Option<f64>), | ||
| /// 32bit decimal, using the i32 to represent the decimal, precision scale | ||
| Decimal32(Option<i32>, u8, i8), |
There was a problem hiding this comment.
I think 🤔 this is technically a breaking API change I think because ScalarValue is a public enum and not parsed non_exhastive... So we can't backport it to the branch-50 without breaking the API
There was a problem hiding this comment.
That makes perfect sense, should've noticed it myself.
I can wait until November, sorry about causing so much noise pushing for this release, at least we'll get the docs fix in.
Which issue does this PR close?
50.2.0(minor) #17849Rationale for this change
Backports initial support for decimal 32/64 to the 50.x branch
What changes are included in this PR?
See original PR:
Are these changes tested?
Includes some tests and reinforced by SLT. Follow up PR contains more tests.
Are there any user-facing changes?
No