I'm trying to package khal 0.8.2 for GNU Guix, but the following test fails: test_set_focus_date_weekstart_6 in tests/ui/test_calendarwidget.py.
Relevant snippet from the build log:
starting phase `check'
============================= test session starts ==============================
platform linux -- Python 3.4.3 -- py-1.4.31 -- pytest-2.7.3
rootdir: /tmp/guix-build-khal-0.8.2.drv-0/khal-0.8.2, inifile:
plugins: cov
collected 176 items
tests/aux_test.py .....................
tests/backend_test.py .........................
tests/cal_display_test.py ....x.
tests/cli_test.py .....x.............
tests/controller_test.py ....
tests/event_test.py .............................
tests/khalendar_aux_test.py ................................
tests/khalendar_test.py ......................
tests/settings_test.py ........
tests/terminal_test.py ...
tests/vtimezone_test.py ...
tests/ui/test_calendarwidget.py ..F
tests/ui/test_widgets.py .
=================================== FAILURES ===================================
_______________________ test_set_focus_date_weekstart_6 ________________________
def test_set_focus_date_weekstart_6():
with freeze_time('2016-04-10'):
today = date.today()
for diff in range(-21, 21, 1):
frame = CalendarWidget(on_date_change=lambda _: None,
keybindings=keybindings,
on_press=on_press,
firstweekday=6,
> weeknumbers='right')
tests/ui/test_calendarwidget.py:48:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
khal/ui/calendarwidget.py:607: in __init__
self.set_focus_date(self._initial)
khal/ui/calendarwidget.py:624: in set_focus_date
self.box.set_focus_date(a_day)
khal/ui/calendarwidget.py:297: in set_focus_date
self.body.set_focus_date(a_day)
khal/ui/calendarwidget.py:371: in set_focus_date
row, column = self.get_date_pos(a_day)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = CalendarWalker([<DateCColumns from 2016-06-26 to 2016-07-02>, <DateCColumns from 2016-07-03 to 2016-07-09>, <DateCColu...4 to 2016-08-20>, <DateCColumns from 2016-08-21 to 2016-08-27>, <DateCColumns from 2016-08-28 to 2016-09-03>], focus=5)
a_day = FakeDate(2016, 4, 10)
def get_date_pos(self, a_day):
"""get row and column where `a_day` is located
:type: a_day: datetime.date
:rtype: tuple(int, int)
"""
# rough estimate of difference in lines, i.e. full weeks, we might be
# off by as much as one week though
week_diff = int((self.focus_date - a_day).days / 7)
new_focus = self.focus - week_diff
# in case new_focus is 1 we will later try set the focus to 0 which
# will lead to an autoprepend which will f*ck up our estimation,
# therefore better autoprepending anyway, even if it might not be
# necessary
if new_focus <= 1:
self._autoprepend()
week_diff = int((self.focus_date - a_day).days / 7)
new_focus = self.focus - week_diff
for offset in [0, -1, 1]: # we might be off by a week
row = new_focus + offset
try:
if row >= len(self):
self._autoextend()
> column = self[row].get_date_column(a_day)
E IndexError: list index out of range
khal/ui/calendarwidget.py:409: IndexError
=============== 1 failed, 173 passed, 2 xfailed in 27.22 seconds ===============
phase `check' failed after 31.5 seconds
In Travis, all the tests in the 0.8.x branch passed, so this certainly could be an issue with dependency versions. From within a Guix development environment (using ./pre-inst-env guix environment khal) git bisect tells me that 8a802825 fixes the bug triggering this test:
8a802825ed3a0f06b387e8af45e68f17cccc0841 is the first fixed commit
commit 8a802825ed3a0f06b387e8af45e68f17cccc0841
Author: Christian Geier <geier@lostpackets.de>
Date: Tue Apr 26 01:00:19 2016 +0200
Fix Calendarwidget: start contstructing inital month
The commit message sounds related to the failed test. I'd rather not package khal with a known bug which is fixed in master, but maybe that's okay. I'd appreciate some guidance on this.
pwnage101
commented
8 years ago
I'm trying to package khal 0.8.2 for GNU Guix, but the following test fails:
test_set_focus_date_weekstart_6intests/ui/test_calendarwidget.py.Relevant snippet from the build log:
In Travis, all the tests in the 0.8.x branch passed, so this certainly could be an issue with dependency versions. From within a Guix development environment (using
./pre-inst-env guix environment khal) git bisect tells me that 8a802825 fixes the bug triggering this test:The commit message sounds related to the failed test. I'd rather not package khal with a known bug which is fixed in master, but maybe that's okay. I'd appreciate some guidance on this.
More boring info below this line:
Full build log from
guix build khal: https://gist.githubusercontent.com/pwnage101/b95dee3e1ff799184a9f6922696f15e8/raw/0b7439a940c4549e54d9242f33fd97f672b30258/build_log.txtHere's a sorted list of all the dependencies being used to build khal and its dependencies, excluding implicit dependencies such as GCC, tar, make, etc.: https://gist.github.com/pwnage101/b95dee3e1ff799184a9f6922696f15e8/raw/0b7439a940c4549e54d9242f33fd97f672b30258/dependency_closure.txt
The guix package definition used for this build: https://gist.githubusercontent.com/pwnage101/b95dee3e1ff799184a9f6922696f15e8/raw/0b7439a940c4549e54d9242f33fd97f672b30258/package_definition.scm