traceback formatting can drop a frame

[benjaminp]

BPO	34588
Nosy	@benjaminp, @serhiy-storchaka, @miss-islington
PRs	closes bpo-34588: Fix an off-by-one error in traceback formatting. #9077 [3.7] bpo-34588: Fix an off-by-one error in traceback formatting. (GH-9077) #9134 [3.6] bpo-34588: Fix an off-by-one error in traceback formatting. (GH-9077) #9135

^{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 2018-09-10.16:01:12.044>
created_at = <Date 2018-09-05.17:52:12.246>
labels = ['interpreter-core', '3.8', 'type-bug', 'library', '3.7']
title = 'traceback formatting can drop a frame'
updated_at = <Date 2018-09-10.16:10:25.501>
user = 'https://github.com/benjaminp'

bugs.python.org fields:

activity = <Date 2018-09-10.16:10:25.501>
actor = 'miss-islington'
assignee = 'none'
closed = True
closed_date = <Date 2018-09-10.16:01:12.044>
closer = 'benjamin.peterson'
components = ['Interpreter Core', 'Library (Lib)']
creation = <Date 2018-09-05.17:52:12.246>
creator = 'benjamin.peterson'
dependencies = []
files = []
hgrepos = []
issue_num = 34588
keywords = ['patch']
message_count = 7.0
messages = ['324648', '324663', '324827', '324858', '324926', '324928', '324931']
nosy_count = 3.0
nosy_names = ['benjamin.peterson', 'serhiy.storchaka', 'miss-islington']
pr_nums = ['9077', '9134', '9135']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue34588'
versions = ['Python 3.6', 'Python 3.7', 'Python 3.8']

[benjaminp]

Consider the following script:
import traceback

def fill_stack(depth):
    if depth <= 1:
        return traceback.format_stack()
    else:
        return fill_stack(depth - 1)

assert fill_stack(4) != fill_stack(5)

On the Python 3 versions I tested, this script doesn't fail! Somehow traceback is producing identical tracebacks for different callstacks.

[benjaminp]

Further investigation reveals this is a general off-by-one error with the recursive traceback pruning feature.

[serhiy-storchaka]

In case of 4 repeated lines it doesn't make much sense to output 3 repeated lines and replace the forth line with "Previous line repeated 1 more time". This doesn't save space and doesn't help reading the traceback. I think it is better to output the forth repeated line.

[benjaminp]

It actually does save space because each traceback entry is usually two lines (frame id and source snippet).

I don't really have an opinion about what should happen be printed on the boundary cases. My current PR seems like a strict improvement to the current implementation, which is just incorrect.

[benjaminp]

New changeset d545869 by Benjamin Peterson in branch 'master':
bpo-34588: Fix an off-by-one error in traceback formatting. (GH-9077)
d545869

[miss-islington]

New changeset 4902017 by Miss Islington (bot) in branch '3.7':
bpo-34588: Fix an off-by-one error in traceback formatting. (GH-9077)
4902017

[miss-islington]

New changeset afb25bc by Miss Islington (bot) in branch '3.6':
bpo-34588: Fix an off-by-one error in traceback formatting. (GH-9077)
afb25bc