settrace docs are wrong about "c_call" events

[nedbat]

BPO	17799
Nosy	@gvanrossum, @nedbat, @zhangyangyu
PRs	bpo-17799: Explain real behaviour of sys.settrace and sys.setprofile #4056 [3.6] bpo-17799: Explain real behaviour of sys.settrace and sys.setprofile (GH-4056) #5298 [2.7] bpo-17799: Explain real behaviour of sys.settrace and sys.setprofile (GH-4056) #5299

^{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 2018-01-24.14:54:45.266>
created_at = <Date 2013-04-20.03:36:20.117>
labels = ['type-bug', '3.7', 'docs']
title = 'settrace docs are wrong about "c_call" events'
updated_at = <Date 2018-01-24.14:54:45.265>
user = 'https://github.com/nedbat'

bugs.python.org fields:

activity = <Date 2018-01-24.14:54:45.265>
actor = 'xiang.zhang'
assignee = 'docs@python'
closed = True
closed_date = <Date 2018-01-24.14:54:45.266>
closer = 'xiang.zhang'
components = ['Documentation']
creation = <Date 2013-04-20.03:36:20.117>
creator = 'nedbat'
dependencies = []
files = []
hgrepos = []
issue_num = 17799
keywords = ['patch']
message_count = 9.0
messages = ['187406', '228143', '309797', '310383', '310393', '310396', '310587', '310598', '310599']
nosy_count = 6.0
nosy_names = ['gvanrossum', 'nedbat', 'ionelmc', 'docs@python', 'raylu', 'xiang.zhang']
pr_nums = ['4056', '5298', '5299']
priority = 'low'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue17799'
versions = ['Python 2.7', 'Python 3.6', 'Python 3.7']

[nedbat]

Looking into this Stack Overflow question: http://stackoverflow.com/questions/16115027/pythons-sys-settrace-wont-create-c-call-events

Reading the code in c_eval.c and friends, it looks like "c_call" events are never passed to the trace function, only to the profile function. The docs are wrong and should be fixed.

The setprofile docs simply point to settrace for details, so the text needs to accommodate both functions' needs.

[BreamoreBoy]

@ned can you provide a patch for this?

[zhangyangyu]

Is there any reason not to respond to "c_call" events in trace function?

[gvanrossum]

The reason not to pass C calls to the tracing function is that tracing exists to support pdb and other debuggers, and pdb only cares about tracing through Python code. So the docs should be updated.

[zhangyangyu]

Hi Guido. Looking at the implementation, it seems pdb ignores c_call not by relying on the interpreter, but bdb takes no action when encountering C events. Yeah, even bdb wrongly expects C events will be triggered for settrace. [1]

And what if some debuggers want not only to trace Python events but also C events, then it has to mix the power of trace and profile but can't only rely on trace.

[1] https://github.com/python/cpython/blob/master/Lib/bdb.py#L59

[gvanrossum]

Too bad, it's been like this since the feature was first created. It would
most likely break other code that uses the tracing hooks, so you'd have to
propose it as a new feature.

[zhangyangyu]

New changeset 131fd7f by Xiang Zhang (Pablo Galindo) in branch 'master':
bpo-17799: Explain real behaviour of sys.settrace and sys.setprofile (bpo-4056)
131fd7f

[zhangyangyu]

New changeset fd844ef by Xiang Zhang in branch '3.6':
bpo-17799: Explain real behaviour of sys.settrace and sys.setprofile (GH-4056) (bpo-5298)
fd844ef

[zhangyangyu]

New changeset e64a47b by Xiang Zhang in branch '2.7':
bpo-17799: Explain real behaviour of sys.settrace and sys.setprofile (GH-4056). (bpo-5299)
e64a47b