strawgate opened a new pull request, #25664:
URL: https://github.com/apache/datafusion/pull/25664

   ## Which issue does this PR close?
   
   Closes #25663.
   
   ## Rationale for this change
   
   Schema adaptation can legitimately encounter columns absent from an older 
physical schema. A missing-column lookup currently formats an error listing 
every physical field, then discards it. This adds avoidable latency as schemas 
widen.
   
   ## What changes are included in this PR?
   
   Use the non-error-producing field lookup in the physical expression adapter. 
Add a Criterion benchmark for present and missing columns in 16- and 128-field 
schemas.
   
   ## What is the testing strategy for this PR?
   
   The existing adapter tests cover missing nullable and non-nullable columns 
and present columns. The full extended workspace test command passed locally, 
as did formatting and strict package-level Clippy. Full-workspace Clippy 
currently reports two warnings in unchanged 
`datafusion/physical-plan/benches/multi_group_by.rs`.
   
   On Apple Silicon, the missing-column benchmark changed from 537 to 256 ns 
for 16 fields and 1.83 us to 302 ns for 128 fields. Present-column timings were 
within run-to-run variation. These are microbenchmarks, not end-to-end query 
measurements.
   
   ## Are there any user-facing changes?
   
   No behavior or API changes; this reduces work for expected missing columns.


-- 
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