andygrove commented on PR #4911:
URL: 
https://github.com/apache/datafusion-comet/pull/4911#issuecomment-5146443923

   @mbutrovich thanks — the test-coverage gap is addressed in 826b2b7f0.
   
   I added your `test_floor_decimal128_wide_array` as written, and verified the 
gap was real by mutating `decimal_floor_wide` to return `div_floor(x, div) + 
1`: all nine pre-existing tests passed against that mutant and only the new 
tests caught it, which is exactly the failure mode you described.
   
   Two adjacent paths were uncovered as well, so I picked those up too:
   
   - **The i64/wide seam.** `test_floor_decimal128_i64_boundary_array` 
straddles `i64::MAX` and `i64::MIN`. Each pair differs by one unscaled unit and 
must floor to the same value, so the 64-bit and 128-bit divisions are asserted 
to *agree* at the switch point in both signs, rather than each being checked in 
isolation.
   - **The unspecialized scale.** `test_floor_decimal128_large_scale_array` 
uses scale 19, where `dispatch_pow10!` stops specializing and falls back to the 
runtime `decimal_floor_f`. That fallback was only ever reached through the 
scalar path.
   
   All 11 floor tests pass, `cargo clippy --all-targets` clean. No production 
code changed, so the benchmark numbers in the PR description still stand.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to