Add a function to get GC statistics

[pitrou]

BPO	16351
Nosy	@gpshead, @jcea, @pitrou, @tiran, @benjaminp, @serhiy-storchaka
Files	gc_get_stats.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 = None
closed_at = <Date 2012-10-30.21:47:16.794>
created_at = <Date 2012-10-28.17:30:20.783>
labels = ['type-feature', 'library']
title = 'Add a function to get GC statistics'
updated_at = <Date 2013-12-26.20:12:54.178>
user = 'https://github.com/pitrou'

bugs.python.org fields:

activity = <Date 2013-12-26.20:12:54.178>
actor = 'python-dev'
assignee = 'none'
closed = True
closed_date = <Date 2012-10-30.21:47:16.794>
closer = 'pitrou'
components = ['Library (Lib)']
creation = <Date 2012-10-28.17:30:20.783>
creator = 'pitrou'
dependencies = []
files = ['27768']
hgrepos = []
issue_num = 16351
keywords = ['patch']
message_count = 11.0
messages = ['174060', '174062', '174063', '174067', '174068', '174069', '174070', '174230', '174231', '174236', '206958']
nosy_count = 7.0
nosy_names = ['gregory.p.smith', 'jcea', 'pitrou', 'christian.heimes', 'benjamin.peterson', 'python-dev', 'serhiy.storchaka']
pr_nums = []
priority = 'low'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue16351'
versions = ['Python 3.4']

[pitrou]

This patch adds a function named gc.get_stats() which returns a list of dictionaries containing per-generation statistics:

>>> import pprint, gc
>>> pprint.pprint(gc.get_stats())
[{'collected': 0, 'collections': 12, 'uncollectable': 0},
 {'collected': 0, 'collections': 1, 'uncollectable': 0},
 {'collected': 0, 'collections': 0, 'uncollectable': 0}]

[tiran]

What are the possible performance implications of the statistics?

[pitrou]

You mean negative implications? None :-)

[serhiy-storchaka]

Why dictionaries and not struct sequences?

[serhiy-storchaka]

Oh, I understand. GC callbacks receive a dict.

For avoid confusion with gc.DEBUG_STATS and to conform with callbacks argument name, may be better to name this function as gc.get_infos([generation])?

[serhiy-storchaka]

I.e. gc.get_info([generation]). Returns the info for specified generation or total if the parameter omitted.

[pitrou]

Why dictionaries and not struct sequences?

Because it's much simpler like that.

For avoid confusion with gc.DEBUG_STATS and to conform with callbacks
argument name, may be better to name this function as
gc.get_infos([generation])?

Well, this is really about statistics, not general information.
Furthermore, it is not related to the callbacks functionality.
As for the generation parameter, I don't think it's useful: the result
list is only 3 elements long.

[python-dev]

New changeset 43d87cdf9457 by Antoine Pitrou in branch 'default':
Issue bpo-16351: New function gc.get_stats() returns per-generation collection statistics.
http://hg.python.org/cpython/rev/43d87cdf9457

[pitrou]

Now committed.

[serhiy-storchaka]

Because it's much simpler like that.

Well, I wrote a patch with structure sequences, it is really much more expansive.

I have some comments.

1. You can allocate list of NUM_GENERATIONS elements and then use PyList_SET_ITEM(result, i, stat). This is 4 lines shorter.

2. And may be return the tuple? get_count() and get_threshold() return tuples.

3. You forgot to add get_stats() to the module docstring.

[python-dev]

New changeset 17bd04fbf3d3 by R David Murray in branch 'default':
whatsnew for gc.get_stats, plus doc tweaks.
http://hg.python.org/cpython/rev/17bd04fbf3d3