This splits a few NixOS tests (namely Chromium, VirtualBox and the
networking tests) into several subtests that are exposed via attributes.
The networking tests were already split up but they didn't expose an
attribute set of available tests but used a function attribute to
specify the resulting test instead.
A new function callSubTests in nixos/release.nix is now responsible for
gathering subtests, which is also used for the installer and boot tests.
The latter is now placed in a tests.boot.* namespace rather than
"polluting" the tests attribute set with its subtest.
As @bobvanderlinden suggests in #13585:
"Looks like that cleans things up quite a bit! Just one aesthetics note,
the boot tests could now be renamed from boot.bootBiosCdrom to
boot.biosCdrom in nixos/tests/boot.nix:L33.
That makes them more consistent with the other tests."
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Some recent perl version introduced "keys" to return the keys
in random order. As some of the packages are solved by "provides" and
based on the order, this randomness affects what packages get into the
closure.
This problem may be in other nix perl scripts.
Updates VirtualBox from version 5.0.12 to 5.0.14.
Upstream changes are (without bug IDs):
* GUI: properly limit the number of VCPUs to the number of physical cores
on Mac OS X
* Audio: fixed a bug which prevented loading a saved state of a saved
guests with HDA emulation (5.0.12 regression)
* Audio: don't crash if the backend is unable to initialize
* Audio: fixed audio capture on Mac OS X
* Storage: fixed a possible crash when attaching the same ISO image
multiple times to the same VM
* BIOS: properly report if two floppy drives are attached
* USB: fixed a problem with filters which would not capture the device
under certain circumstances (5.0.10 regression)
* ExtPack: black-list Extension Packs older than 4.3.30 due to
incompatible changes not being properly handled in the past
* Windows hosts: fixed a regression which caused robocopy to fail
* Linux hosts: properly create the /sbin/rcvboxdrv symbolic link (5.0.12
regression)
* Mac OS X hosts: several fixes for USB on El Capitan
* Linux Additions: fixes for Linux 4.5
Full upstream changelog with bug IDs can be found at:
https://www.virtualbox.org/wiki/Changelog
The reason I was reluctant to merge this before were these symbol lookup
errors:
vboxsf: Unknown symbol VBoxGuest_RTMemTmpFree (err 0)
vboxsf: Unknown symbol VBoxGuestIDCCall (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemFastMutexRequest (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemFastMutexRelease (err 0)
vboxsf: Unknown symbol VBoxGuest_RTLogRelGetDefaultInstanceEx (err 0)
vboxsf: Unknown symbol VBoxGuest_RTErrConvertToErrno (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemFastMutexCreate (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemFastMutexDestroy (err 0)
vboxsf: Unknown symbol VBoxGuest_RTMemContFree (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemMutexRelease (err 0)
vboxsf: Unknown symbol VBoxGuestIDCOpen (err 0)
vboxsf: Unknown symbol VBoxGuest_RTAssertShouldPanic (err 0)
vboxsf: Unknown symbol VBoxGuest_RTMemContAlloc (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemMutexRequest (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemMutexCreate (err 0)
vboxsf: Unknown symbol VBoxGuest_RTMemTmpAllocTag (err 0)
vboxsf: Unknown symbol VBoxGuest_RTSemMutexDestroy (err 0)
vboxsf: Unknown symbol VBoxGuest_RTAssertMsg1Weak (err 0)
vboxsf: Unknown symbol VBoxGuestIDCClose (err 0)
vboxsf: Unknown symbol VBoxGuest_RTAssertMsg2Weak (err 0)
However, after testing it against 5.0.12, the same errors occur there as
well, so it is likely related to our VM tests.
This makes it easier to test just a specific channel rather than to
force testing all builds down the users/testers throat. Especially this
makes it easier to test NixOS channel upgrades only against the Chromium
stable channel instead of just removing the beta/dev channels from the
tests entirely (as done in 69ec09f38a).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Should clean up a lot of these redundant lines for various sub-tests.
Note that the tests.boot* are now called tests.boot.boot*, but otherwise
all the test attribute names should stay the same.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Cc: @edolstra
Cc: @wkennington
Cc: @bobvanderlinden
So far the networking test expression only generated a single test
depending on the passed "test" attribute. This makes it difficult to
autodiscover the subtests with our shiny new callSubTests function.
This change essentially doesn't change the behaviour of the subtests but
rather exposes them as an attribute set instead of relying on a
particular input argument.
The useNetworkd argument still exists however.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Cc: @wkennington
This should de-clutter the various redundant lines of callTest's on
subtests so that every main test file should have only one line with a
callSubTests instead.
Overrides work the same way as callTest, except that if the system
attribute is explicitly specified we do not generate attributes for all
available systems.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Cc: @edolstra
Otherwise, a call to, for example, nix-prefetch-git, in a nix-shell --pure or
nix-build phase would result in an error such as:
Switched to a new branch 'fetchgit'
removing `.git'...
/nix/store/7qvjji7kbd033nsyxlpiiiam61a44yz9-nix-prefetch-git/bin/.nix-prefetch-git-wrapped: line 374: nix-hash: command not found
