audioop functions shouldn't accept strings

[serhiy-storchaka]

BPO	16685
Nosy	@jcea, @jaraco, @vstinner, @ezio-melotti, @serhiy-storchaka
Dependencies	bpo-16682: Document that audioop works with bytes, not strings
Files	audioop_buffer.patch audioop_buffer_2.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 2013-11-11.17:54:29.465>
created_at = <Date 2012-12-14.17:37:35.570>
labels = ['extension-modules', 'easy', 'type-feature']
title = "audioop functions shouldn't accept strings"
updated_at = <Date 2014-03-10.01:35:02.697>
user = 'https://github.com/serhiy-storchaka'

bugs.python.org fields:

activity = <Date 2014-03-10.01:35:02.697>
actor = 'python-dev'
assignee = 'serhiy.storchaka'
closed = True
closed_date = <Date 2013-11-11.17:54:29.465>
closer = 'serhiy.storchaka'
components = ['Extension Modules']
creation = <Date 2012-12-14.17:37:35.570>
creator = 'serhiy.storchaka'
dependencies = ['16682']
files = ['32241', '32276']
hgrepos = []
issue_num = 16685
keywords = ['patch', 'easy']
message_count = 18.0
messages = ['177477', '177487', '177488', '177513', '177515', '177516', '177517', '177522', '200556', '200675', '200738', '202499', '202560', '202596', '202606', '202635', '203030', '213004']
nosy_count = 6.0
nosy_names = ['jcea', 'jaraco', 'vstinner', 'ezio.melotti', 'python-dev', 'serhiy.storchaka']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue16685'
versions = ['Python 3.4']

[serhiy-storchaka]

Inadvertently strings accepted as arguments of audioop functions. This is a meaningless behavior and remnant of Python 2. We should drop string support.

[ezio-melotti]

What happens if you pass strings?
If it doesn't work we can just fix it to raise an exception and stop returning random results, if it works but bytes should be used instead, we should go through a deprecation process.

[serhiy-storchaka]

What happens if you pass strings?

They are encoded with UTF-8. See 's#' and 's*' formats in PyArg_ParseTuple() ('y*' recommended for bytes).

[vstinner]

Accepting Unicode strings is surprising and must fail. I don't think that a deprecation process is required here.

[serhiy-storchaka]

Accepting Unicode strings is surprising and must fail. I don't think that a deprecation process is required here.

Should it be fixed in 3.4 only or in all 3.x?

[ezio-melotti]

The deprecation process can be avoided only if passing strings results in a meaningless result. If there are cases where users are passing strings and everything works fine -- even if they should be passing bytes instead -- we should deprecate strings first to warn the users that the "feature" is going to be removed.

[serhiy-storchaka]

In 2.x it was have a little sense (if you use some 8-bit encoding as default encoding, but the default was 7-bit ascii). But with utf-8 it doesn't have sense and should quickly lead to an unexpected result (or just fail in most cases).

[vstinner]

Should it be fixed in 3.4 only or in all 3.x?

Such change cannot be done in a minor release (like 3.3.x), so it can
only be done in Python 3.4.

[serhiy-storchaka]

Here is a patch. Now audioop functions no more accept str, but accept bytes-like objects instead.

[vstinner]

audioop_buffer.patch: In general, I don't like goto in C. But to cleanup something when exiting a function, I like goto to implement something similar to Python try/finally. What do you think Serhiy?

[serhiy-storchaka]

Well. In updated patch I use gotos where it decreases the number of lines (total decreasing is almost 30 lines).

[serhiy-storchaka]

If there are no objections I will commit last patch tomorrow.

[python-dev]

New changeset bab0cbf86835 by Serhiy Storchaka in branch 'default':
Issue bpo-16685: Added support for any bytes-like objects in the audioop module.
http://hg.python.org/cpython/rev/bab0cbf86835

[jaraco]

The patch as committed causes the Windows 64-bit builds to fail to compile. http://buildbot.python.org/all/builders/AMD64%20Windows7%20SP1%203.x/builds/3187/steps/compile/logs/stdio

[python-dev]

New changeset 35cd00465624 by Serhiy Storchaka in branch 'default':
Fixed compile error on Windows caused by arithmetic with void * pointers
http://hg.python.org/cpython/rev/35cd00465624

[serhiy-storchaka]

Fixed. Thank you Jason.

[python-dev]

New changeset b96f4ee1b08b by Serhiy Storchaka in branch 'default':
Issue bpo-16685: Added support for writing any bytes-like objects in the aifc,
http://hg.python.org/cpython/rev/b96f4ee1b08b

[python-dev]

New changeset c375697f062e by R David Murray in branch 'default':
whatsnew: auidioop.byteswap (bpo-16685).
http://hg.python.org/cpython/rev/c375697f062e