optimization: IPC: pass FieldNodes around by value instead of reference#543

Open

pixelherodev wants to merge 28 commits intoapache:mainfrom

pixelherodev:avoid-allocations

Commits on Oct 23, 2025

optimization: IPC: pass FieldNodes around by value instead of reference
Noam Preil
committed
optimization: compare: avoid reflection in union comparison
Noam Preil
committed
cookie test: use maps.Equal instead of reflect.DeepEquals
Noam Preil
committed
cleanup: internal/flatbuf/message: embed flatbuffers.Table directly
Noam Preil
committed
internal/flatbuf/Message: embed Message in parent allocation
Noam Preil
committed
cleanup: minSpaceSavings need not have a pointer
Noam Preil
committed
optimization: NewReader: reuse the same Config
Noam Preil
committed
optimization: allocator: remove unnecessary branch
Noam Preil
committed
optimization: MessageReader: use a static 4-byte buffer for header reads
Noam Preil
committed
cleanup: just use newConfig(opts...) instead of reapplying
Noam Preil
committed

Commits on Oct 24, 2025

optimization: IPC: allows inlining of schema()
Noam Preil
committed
optimization: don't allow flatbuf-initialized types to escape to heap
Noam Preil
committed
optimization: ipc/Message: reuse the same *memory.Buffer for meta
Noam Preil
committed
optimization: ipc/message: reuse *Buffer for message body
Noam Preil
committed
cleanup: preallocate meta Buffer
Noam Preil
committed
cleanup: buffer: detect resize to zero earlier
Noam Preil
committed
flatbuf.RecordBatch should not escape to heap
Noam Preil
committed
minor cleanup: modernize benchmarking loop
Noam Preil
committed
refactor msg-header handling: drop unnecessary wrappers and abstraction
Noam Preil
committed
minor cleanup
Noam Preil
committed
internal/flatbuf: cleanup Table wrapping
Noam Preil
committed
cleanup flatbuf initialization systematically
Noam Preil
committed
minor cleanup to readDictionary: pass Message instead of components
Noam Preil
committed
Regenerate flatbuf with an optimized flatc, update usage
Noam Preil
committed
lint
Noam Preil
committed
restore licenses
Noam Preil
committed
regen flatbufs with more cleanup, update usage
Noam Preil
committed
drop another allocation per message where possible
Noam Preil
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

optimization: IPC: pass FieldNodes around by value instead of reference#543

optimization: IPC: pass FieldNodes around by value instead of reference#543
pixelherodev wants to merge 28 commits intoapache:mainfrom
pixelherodev:avoid-allocations

Commits on Oct 23, 2025

optimization: IPC: pass FieldNodes around by value instead of reference

optimization: compare: avoid reflection in union comparison

cookie test: use maps.Equal instead of reflect.DeepEquals

cleanup: internal/flatbuf/message: embed flatbuffers.Table directly

internal/flatbuf/Message: embed Message in parent allocation

cleanup: minSpaceSavings need not have a pointer

optimization: NewReader: reuse the same Config

optimization: allocator: remove unnecessary branch

optimization: MessageReader: use a static 4-byte buffer for header reads

cleanup: just use newConfig(opts...) instead of reapplying

Commits on Oct 24, 2025

optimization: IPC: allows inlining of schema()

optimization: don't allow flatbuf-initialized types to escape to heap

optimization: ipc/Message: reuse the same *memory.Buffer for meta

optimization: ipc/message: reuse *Buffer for message body

cleanup: preallocate meta Buffer

cleanup: buffer: detect resize to zero earlier

flatbuf.RecordBatch should not escape to heap

minor cleanup: modernize benchmarking loop

refactor msg-header handling: drop unnecessary wrappers and abstraction

minor cleanup

internal/flatbuf: cleanup Table wrapping

cleanup flatbuf initialization systematically

minor cleanup to readDictionary: pass Message instead of components

Regenerate flatbuf with an optimized flatc, update usage

lint

restore licenses

regen flatbufs with more cleanup, update usage

drop another allocation per message where possible

Uh oh!

optimization: IPC: pass FieldNodes around by value instead of reference#543pixelherodev wants to merge 28 commits intoapache:mainapache/arrow-go:mainfrom pixelherodev:avoid-allocationspixelherodev/arrow-go:avoid-allocationsCopy head branch name to clipboard

Commits

Commits on Oct 23, 2025

Commits on Oct 24, 2025

optimization: IPC: pass FieldNodes around by value instead of reference#543
pixelherodev wants to merge 28 commits intoapache:mainfrom
pixelherodev:avoid-allocations