Rich-T-kid opened a new pull request, #12252:
URL: https://github.com/apache/arrow-rs/pull/12252

   # Which issue does this PR close?
   
   Relates to #12210.
   
   # Rationale for this change
   
   The single-buffer indexed ByteView copy path added in #12212 still converts 
the destination length to `u32` for every selected non-inline value. The final 
buffer length is already bounded before the loop, so the repeated overflow 
branch is unnecessary.
   
   # What changes are included in this PR?
   
   This checks once that `max_buffer_size` fits in a ByteView offset. The 
common copy loop then uses the existing final-buffer-length bound to write each 
destination offset without a repeated checked conversion.
   
   # Are these changes tested?
   
   The existing exact-fit and split-buffer tests cover both copy paths. In 
local A/B measurements against #12212, four Utf8View/BinaryView cases improved 
by 1.6-4.1% and the remaining four were statistically neutral, with no 
regressions. Hosted results are pending.
   
   # Are there any significant user-facing changes?
   
   No.
   


-- 
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]

Reply via email to