summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* ASP: suggest to connect in install failurebugfix/16566-asp-dl-failure-case-C-with-notificationAlan2019-03-161-1/+2
| | | | Will-fix: #16566
* ASP: ignore install exit codeAlan2019-03-161-1/+1
| | | | | | We want to run the upgrade even if the install step fails, so let's ignore its exit code. Will-fix: #15957
* Merge branch 'bugfix/16559-tor-browser-8.0.7+force-all-tests' into stable ↵intrigeri2019-03-152-26/+26
|\ | | | | | | (Fix-committed: #16559)
| * Fetch Tor Browser from our own archive.intrigeri2019-03-141-1/+1
| |
| * Upgrade Tor Browser to 8.0.7-build3.intrigeri2019-03-142-26/+26
| |
* | Merge branch 'bugfix/16452-remove-useless-extra-initramfs-update' into ↵intrigeri2019-03-152-7/+30
|\ \ | |/ |/| | | stable (Fix-committed: #16452)
| * Sanity check the size of the initramfs (refs: #16452)intrigeri2019-03-131-0/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This will help us detect if any of this happens: - config/chroot_local-hooks/99-initramfs-compress is broken - live-build does not generate the initramfs after config/chroot_local-hooks/99-initramfs-compress has applied our preferred configuration - Some unrelated change makes the initramfs substantially larger, as an unintended side-effect. As of Tails 3.12.1, our current initramfs is 30M large so a 35M limit should give us just enough safety margin.
| * Drop useless manual initramfs update (refs: #16452)intrigeri2019-03-131-7/+5
| | | | | | | | | | | | live-build will do that itself later on. This saves ~2 minutes (~3%) on the total build time.
* | Merge remote-tracking branch 'origin/bugfix/16348-tor-0.3.5+force-all-tests' ↵intrigeri2019-03-142-6/+1
|\ \ | |/ |/| | | into stable (Fix-committed: #16348)
| * Bump APT snapshot of the torproject archive to 2019031301 (refs: #16348).intrigeri2019-03-131-1/+1
| | | | | | | | The 2019022601 snapshot has expired and was garbage-collected.
| * Revert "Install tor 0.3.4 from out custom APT repo (refs: #16348, #16349)."Sandro Knauß2019-03-131-5/+0
| | | | | | | | This reverts commit 3d88e030ca9aac8348b46d08c4994c19b2bb607e.
| * Bump torproject/serial to 2019022601.Sandro Knauß2019-03-131-1/+1
|/
* Merge branch 'bugfix/16338-tor-launcher-localization+force-all-tests' into ↵intrigeri2019-03-111-4/+9
|\ | | | | | | stable (Fix-committed: #16338)
| * Update comment to match current code (refs: #16338)intrigeri2019-03-111-2/+4
| | | | | | | | | | commit:7dc8f638eaf00201a441d5a74337174c1d3912b7 introduced usage of $TBB_EXT.
| * Tor Launcher: add langpacks to enable localization (Will-fix: #16338).anonym2019-02-221-2/+5
| | | | | | | | | | Regression introduced when migrating to Tor Browser 8.x in Tails 3.9~rc1 (Tails 3.8 with the last Tor Browser 7.5.x release is fine).
* | Merge branch 'test/14596-automated-tests-for-ASP-gui' into stable (refs: #14596)intrigeri2019-03-1112-101/+397
|\ \
| * | Test suite: fix expected window title (refs: #14596)intrigeri2019-03-061-1/+1
| | |
| * | Test suite: use here-document for readability (refs: #14596)intrigeri2019-03-051-2/+4
| | |
| * | Test suite: make function names clearer and less generic (refs: #14596)intrigeri2019-03-052-6/+6
| | |
| * | Test suite: drop unused variable (refs: #14596)intrigeri2019-03-051-1/+1
| | |
| * | Test suite: use $vm.execute vs $vm.execute_successfully consistently (refs: ↵intrigeri2019-03-052-4/+8
| | | | | | | | | | | | | | | | | | | | | #14596) … and document a bit a few cases where one might wonder why we're not using vm.execute_successfully.
| * | Test suite: fix buggy checks (refs: #14596)intrigeri2019-03-051-2/+2
| | | | | | | | | | | | They were previously no-ops as we did not act based on their return value.
| * | Test suite: fix grammar (refs: #14596)intrigeri2019-03-052-5/+5
| | |
| * | Test suite: use the same terminology as what we expose to users (refs: #14596)intrigeri2019-03-052-4/+4
| | | | | | | | | | | | That's kinda what BDD is about :)
| * | Test suite: add comments about disks life-cycle (refs: #14596)intrigeri2019-03-052-0/+4
| | |
| * | Test suite: replace workaround for bug that's been fixed with proper ↵intrigeri2019-03-052-6/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | implementation (refs: #14596) commit:d2a67e7f7bcafdd22186a91f3b076a01797d1107 added the workaround and documented the bug but no ticket was filed back then. Thankfully anonym's review highlighted this problem as a probable bug in the code, which turned out to be the case; finally, the bug (#16475) was fixed.
| * | Merge remote-tracking branch 'origin/stable' into ↵intrigeri2019-03-046-9/+27
| |\ \ | | | | | | | | | | | | test/14596-automated-tests-for-ASP-gui
| * | | Test suite: update button label to match fixed implementation (refs: #14596, ↵intrigeri2019-03-041-1/+1
| | | | | | | | | | | | | | | | #16110)
| * | | Merge remote-tracking branch 'origin/stable' into ↵intrigeri2019-02-21165-5923/+8501
| |\ \ \ | | | |/ | | |/| | | | | test/14596-automated-tests-for-ASP-gui
| * | | Merge remote-tracking branch 'origin/stable' into ↵intrigeri2019-02-11159-7017/+10093
| |\ \ \ | | | | | | | | | | | | | | | test/14596-automated-tests-for-ASP-gui
| * | | | Test suite: don't rely on the state of the APT lists to check if a package ↵intrigeri2019-01-271-3/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | is installed (refs: #14596) I've seen cases where "apt-cache policy" would fail, when run by 'the package "cowsay" is not installed', as part of "I am notified when Additional Software fails to install a package", with: E: Malformed Description-md5 line; doesn't have the required length (32 != 29) '0143a1c3acbdb045e4fcaab0d8657" This seems wrong and I don't know why this happens, but let's not get side-tracked: the goal here is to check whether a given package is installed, not to validate the state of the APT lists. So let's use dpkg instead. For details about why "dpkg -s" was initially replaced with "apt-cache policy", see c77c2dd90480f8eb53e0c7861e531574345659ec. The analysis of the problem in that commit message was correct: "deinstall" means "The package is selected for deinstallation" and is indeed a transient state. The solution did work until we face a situation when the higher level tool fails while checking things we're not interested in.
| * | | | Test suite: try to avoid race condition (refs: #14596)intrigeri2019-01-272-1/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | My commit fd82498abd1edf7d67bfcde035abe72152a96fb6 was overly optimistic: the race condition wrt. catching the "The installation of your additional software failed" notification was not _solely_ caused by the bug I've fixed in 682ea4607780b41a8b94ada84d9b709e2d746b6f. Additionally, sometimes "all notifications have disappeared" (which actively clears the list of notifications) hides the very notification we're looking for and want to interact with. Whenever this happens, 'I see the "The installation of your additional software failed" notification after at most 300 seconds' will succeed anyway (Dogtail will find notifications even after we've cleared them, go figure) but thankfully 'I can open the Additional Software log file from the notification' will fail to interact with an invisible notification (anything else would be utterly confusing). To avoid this problem, we need to stop clearing all notifications. But then the "Warning: virtual machine detected!" notification may hide the one we're looking for. So let's disable the service that would display this spurious notification. Given this scenario runs Tails offline, this should be the only notification displayed at login time, so this blocking it should be a sufficient substitute for clearing all notifications.
| * | | | Test suite: document potential reliability issue (refs: #14596)intrigeri2019-01-271-0/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | I doubt we'll lose this race in practice so I won't block on this. But it's worth documenting, in order to save some time to whoever debugs potential reliability problems here in the future.
| * | | | Test suite: don't look for an unrelated notification when checking if ↵intrigeri2019-01-271-1/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | tails-additional-software-upgrade.service has started (refs: #14596) This service does not display notifications on success: only tails-additional-software-install.service does. Context: I'm trying to wrap my mind around how Additional Software notifications are handled in the test suite and it confused me a bit that we were looking for a notification that is unrelated to what this step is testing.
| * | | | Test suite: fix comment (refs: #14596)intrigeri2019-01-271-1/+1
| | | | | | | | | | | | | | | | | | | | Currently _all_ following scenarios reuse the "__internal" drive.
| * | | | Test suite: fix erroneous inter-scenario dependency doc (refs: #14596)intrigeri2019-01-271-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | "I set up Additional Software when installing a package without persistent partition and the package is installed next time I start Tails" is not sufficient here: it does not add cowsay to the list of Additional Software. So let's instead declare a dependency on the only scenario that leaves cowsay in this list once it completes.
| * | | | Test suite: make scenario title match what it's actually testing (refs: #14596)intrigeri2019-01-261-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | This scenario does not actually check whether a notification is displayed so let's not pretend it does.
| * | | | Test suite: fix documentation.intrigeri2019-01-261-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | Drives created this way are *not* deleted at the end of the feature by default. AFAICT they've never been.
| * | | | Test suite: clarify comment (refs: #14596)intrigeri2019-01-261-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - All snapshots used by this feature are temporary except "with-network-logged-in" and a couple other snapshots that almost all features use anyway. - It's not relevant whether a disk created by this feature can be reused: on the contrary, here we're trying to prioritize features that use large amounts of disk space _temporarily_. So let's not suggest the opposite. Besides, if the __internal disk is actually reused as-is by following features, we have a problem: every such feature would have Additional Software enabled. Hopefully this is not how our QCOW2 snapshots mechanism works (I did not check yet). - This mechanism is about disk usage in /tmp/TailsToaster, not about memory usage. The situation when a tmpfs is mounted on /tmp/TailsToaster is a particular case, rather than something we can assume to be the default here.
| * | | | Test suite: add missing file extension to picture (refs: #14596)intrigeri2019-01-251-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Sikuli does not mind and will find the file even if we forget to make the extension explicit. But our mechanism to save fuzzy matching candidates won't add the extension itself, so it would save a "ASPDocumentationInstallCloning" file as candidate, instead of the expected "ASPDocumentationInstallCloning.png".
| * | | | Test suite: add comment to make this scenario's secondary responsibility ↵intrigeri2019-01-251-0/+2
| | | | | | | | | | | | | | | | | | | | clear (refs: #14596)
| * | | | Test suite: remove now unused snapshot (refs: #14596)intrigeri2019-01-251-13/+0
| | | | | | | | | | | | | | | | | | | | | | | | | Commit caad88e0ef073bb56760a272bd38417ee7f35ec0 removed the only user of this short-lived snapshot.
| * | | | Test suite: remove workaround for race condition that's been fixed (refs: ↵intrigeri2019-01-252-18/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | #14596) It's likely that "Trying to catch the notification at desktop startup is racy" was only due to us not actually disabling the service we meant to disable, which was fixed by 682ea4607780b41a8b94ada84d9b709e2d746b6f; so hopefully this workaround is not needed anymore.
| * | | | Test suite: update outdated picture (refs: #14596)intrigeri2019-01-251-0/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The phrasing of the GUI has changed in t-p-s.git's commit 41252b861b2d3366a794d8d22808a067f9c1f19e, last April 27. Since then this test has kept passing only because of "Found fuzzy candidate picture for ASPPersistenceSetupOptionEnabled with similarity 0.8", which is actually a bug (fuzzy matching is supposed to be disabled by default in our test suite but for some reason, at least in some cases, it's enabled when it should not). So let's replace this picture with a current one.
| * | | | Test suite: use consistent terminology (refs: #14596)intrigeri2019-01-252-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | We have a few other similar steps where we don't mention the "link". Let's keep this simple.
| * | | | Revert "Test suite: reintroduce check in ASP Debian package removal." (refs: ↵intrigeri2019-01-251-3/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | #14596) This reverts commit 19a6c781ac399c640778f7ed0ddf00b199ff659a. This workaround was only needed due to a race condition caused by the test suite bug which the previous commit fixes.
| * | | | Test suite: actually disable tails-additional-software-install.service ↵intrigeri2019-01-252-6/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | (refs: #14596) The system-wide tails-additional-software-install.service is not "enabled" in Tails: instead it's started by the corresponding user unit (/usr/lib/systemd/user/tails-additional-software-install.service). So the previous call to "systemctl disable" was a no-op. Let's instead disable the user service. And then there's nothing left for us to (re-)enable: we can as well just start the system-wide service ourselves.
| * | | | Merge remote-tracking branch 'origin/testing' into ↵intrigeri2019-01-25753-37415/+31020
| |\ \ \ \ | | | | | | | | | | | | | | | | | | test/14596-automated-tests-for-ASP-gui
| * \ \ \ \ Merge remote-tracking branch 'origin/stable' into ↵Alan2018-12-21352-7934/+13943
| |\ \ \ \ \ | | | | | | | | | | | | | | | | | | | | | feature/14596-automated-tests-for-ASP-gui-on-stable
| * \ \ \ \ \ Merge branch 'stable' into feature/14596-automated-tests-for-ASP-gui-on-stableintrigeri2018-11-278-52/+437
| |\ \ \ \ \ \