This bug was fixed in the package online-accounts-api -
0.1+16.10.20160722.4-0ubuntu1
---
online-accounts-api (0.1+16.10.20160722.4-0ubuntu1) yakkety; urgency=medium
[ Alberto Mardegan ]
* Tests: keep the service alive while running the test (LP: #1603706)
* Enable coverage
For the record, I've found the cause of the issue I wrote about in
comment 4: the Accounts::Manager object was not being deleted at the end
of the test. Given that the D-Bus connection was being torn down, there
probably was some GLib event about that left in the mainloop, which was
processed only
I'm trying to fix this as discussed in IRC, that is respawning the D-Bus
test connection (and related mock services) at every test, and having a
longer timeout.
However I'm getting a crash when calling startServices():
http://paste.ubuntu.com/20160205/
The crash seems to happen after one
With the attached branch added to the silo, we got failures for 2 out of
the 20 builds.
First was the vivid i386 build: https://launchpadlibrarian.net/273878957
/buildlog_ubuntu-vivid-i386.online-accounts-
api_0.1+15.04.20160719-0ubuntu1_BUILDING.txt.gz
1468915489.236
** Branch linked: lp:~mardy/online-accounts-api/timeout-1603706
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1603706
Title:
Possible races in tst_daemon tests.
To manage notifications about this
I've spent some time to investigate this bug, here's my understanding.
The inactivity timer operates on a property of the OA.Manager object,
called "idle". If "idle" becomes true and stays true for the whole
duration of the timeout period (which by default is 5 seconds, and in
the tests is set to
I definitely see how these failures can happen: the tests set the
OnlineAccounts.Manager service's auto-termination timeout to 2 seconds,
which is generally appropriate. However, the testAccountChanges() test
is not even trying to keep the service alive, because it's modifying the
accounts DB via
