zeroshade commented on issue #863:
URL: https://github.com/apache/arrow-go/issues/863#issuecomment-5501435809

   Closing this as not reproducible, based on the analysis in 
https://github.com/apache/arrow-go/issues/863#issuecomment-4838489974 and the 
follow-up.
   
   To recap the two points, since neither has been contested:
   
   1. The attached reproducer sets `BloomFilterOffset = 0`, so 
`GetColumnBloomFilter` returns `nil, nil` before it reads the header, allocates 
a bitset, touches the pool, or registers cleanup. The bloom filter pool path is 
never exercised by that test.
   2. `ResizeNoShrink(0)` does not destroy capacity — it deliberately keeps the 
capacity and resets only the length, so a later resize of a recycled buffer 
does not realloc. This is still the behavior on main 
(`parquet/metadata/cleanup_bloom_filter.go`).
   
   The reported `TotalAlloc` figure is cumulative allocation churn rather than 
retained memory, so on its own it isn't evidence of pool pollution.
   
   That said, I don't want to dismiss a real problem if there is one under 
this. If you can share a reproducer that actually reaches the bloom filter read 
path (non-zero `BloomFilterOffset`) and still shows the degradation, please 
reopen or file a new issue and I'll take another look.


-- 
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