bytes(x) is slow when x is bytearray

[methane]

BPO	27704
Nosy	@vstinner, @methane, @serhiy-storchaka, @matrixise
Files	fast-bytearray-fromobject.patch fast-bytearray-fromobject.patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/serhiy-storchaka'
closed_at = <Date 2016-08-15.08:55:08.607>
created_at = <Date 2016-08-07.19:22:27.381>
labels = ['interpreter-core', 'performance']
title = 'bytes(x) is slow when x is bytearray'
updated_at = <Date 2016-08-15.08:55:51.642>
user = 'https://github.com/methane'

bugs.python.org fields:

activity = <Date 2016-08-15.08:55:51.642>
actor = 'serhiy.storchaka'
assignee = 'serhiy.storchaka'
closed = True
closed_date = <Date 2016-08-15.08:55:08.607>
closer = 'serhiy.storchaka'
components = ['Interpreter Core']
creation = <Date 2016-08-07.19:22:27.381>
creator = 'methane'
dependencies = []
files = ['44039', '44059']
hgrepos = []
issue_num = 27704
keywords = ['patch']
message_count = 8.0
messages = ['272130', '272131', '272249', '272258', '272260', '272261', '272721', '272739']
nosy_count = 5.0
nosy_names = ['vstinner', 'methane', 'python-dev', 'serhiy.storchaka', 'matrixise']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue27704'
versions = ['Python 3.6']

[methane]

When bytes(x), bytes_new checks if x is integer via PyNumber_AsSize_t(x).
It cause TypeError internally.

When x is not an integer, especially bytearray or memoryview, the internal
exception cause significant overhead.

# HEAD
$ ./python -m timeit -s 'data=bytearray(b"xyz")' 'bytes(data)'
1000000 loops, best of 3: 0.696 usec per loop
$ ./python -m timeit -s 'data=bytearray(b"xyz")' 'bytes(data)'
1000000 loops, best of 3: 0.699 usec per loop
$ ./python -m timeit -s 'data=bytearray(b"xyz")' 'bytes(data)'
1000000 loops, best of 3: 0.701 usec per loop

# this patch
$ ./python -m timeit -s 'data=bytearray(b"xyz")' 'bytes(data)'
1000000 loops, best of 3: 0.265 usec per loop
$ ./python -m timeit -s 'data=bytearray(b"xyz")' 'bytes(data)'
1000000 loops, best of 3: 0.265 usec per loop
$ ./python -m timeit -s 'data=bytearray(b"xyz")' 'bytes(data)'
1000000 loops, best of 3: 0.267 usec per loop

[matrixise]

the patch seems to be fine.

[serhiy-storchaka]

bytearray suffers from the same issue. It would be nice to optimize it too.

[matrixise]

Hi Serhiy

Thank you for your feedback.

Stephane

[methane]

Thanks for comments.

[serhiy-storchaka]

LGTM.

[python-dev]

New changeset 789a42401009 by Serhiy Storchaka in branch 'default':
Issue bpo-27704: Optimized creating bytes and bytearray from byte-like objects
https://hg.python.org/cpython/rev/789a42401009

[serhiy-storchaka]

Thank you for your contribution Naoki.