This change allows users to specify an alternative database method. For
example an mpd satellite setup where another mpd on the network shares
it's database with the local instance. The `dbFile` parameter must not be
configured in that case.
Note: I'm not packaging this as stable for now as there might be some
issues left (see GitHub), it says "Version 1.3 alpha" in the bottom left
corner, and I've only got a "New in version 1.2.25" message.
Removed some redundancy (src check via meta.platforms) and made some
changes according to our style-guide.
I've changed meta.description and added meta.longDescription.
If executed in a pure environment (nix-shell --pure) or depending on the
`gtk3` version of the system Signal-Desktop was e.g. crashing when
clicking on a PDF attachment (instead of showing the dialog to save a
file).
Using wrapGAppsHook and setting XDG_DATA_DIRS to the correct version
fixes this bug.
The error message was the following:
```
(signal-desktop:30756): Gtk-WARNING **: 14:04:49.073: Could not find the icon 'user-home-symbolic-ltr'. The 'hicolor' theme
was not found either, perhaps you need to install it.
You can get a copy from:
http://icon-theme.freedesktop.org/releases
(signal-desktop:30756): GLib-GIO-ERROR **: 14:04:49.134: No GSettings schemas are installed on the system
Trace/breakpoint trap
```
Also remove the upstream device trees beforehand so we don't
accidentally end using them if we mess up in postFixup.
Also on the topic of mess ups, make the copy unconditional since they
always seem to be built on all architectures anyway so checking for
their existence just hides typos.
While building the container there are a few occasions where stdenvNoCC
is used underneath. During the last staging merge, some change now tries
to build texinfo during the test while building stdenvNoCC.
With this change, I'm adding stdenvNoCC to the closure to make sure that
even when we have future stdenv changes, it doesn't break (well, except
if we do have another variation like stdenvNoCC that overrides stdenv).
I haven't bisected the exact change, but I'd suspect that it could be
one of the commits in #39457.
This fixes the test and it no longer fails with the following error:
error: unable to download 'http://ftpmirror.gnu.org/texinfo/texinfo-6.5.tar.xz': Couldn't resolve host name (6)
builder for '/nix/store/r7sf1wjbnimwgnv276jh59nfnzw40x30-texinfo-6.5.tar.xz.drv' failed with exit code 1
cannot build derivation '/nix/store/5w1pv788ayi1wahyy76i90yqv96ai4h5-texinfo-6.5.drv': 1 dependencies couldn't be built
cannot build derivation '/nix/store/cnsfkf0j5xmm14zzm5a3a66pz66gbc82-stdenv-linux.drv': 1 dependencies couldn't be built
cannot build derivation '/nix/store/11kkhk57ic8kfd7g197sqwgd0pzqfjhl-nixos-system-foo-0-18.09pre-git.drv': 1 dependencies couldn't be built
error: build of '/nix/store/11kkhk57ic8kfd7g197sqwgd0pzqfjhl-nixos-system-foo-0-18.09pre-git.drv' failed
/run/current-system/sw/bin/nixos-container: failed to build initial container configuration
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @aristidb, @edolstra, @chaoflow, @kampfschlaefer
When -O2 from hardening does not redefine -O3 from CMake, the build fails with:
src/qpid/broker/SelectorExpression.cpp: In member function ‘qpid::broker::Expression* qpid::broker::Parse::orExpression(qpid::broker::Tokeniser&)’:
src/qpid/broker/SelectorExpression.cpp:1041:13: error: ‘*((void*)& s +17)’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (s[1]=='b' || s[1]=='B') {
nix-repl is not needed anymore because Nix 2.0 comes with repl already. Anyway nix-repl is broken right now on Hydra because Nix 1.11 is broken.
https://hydra.nixos.org/build/74933091
Applies a patch to the dd-agent derivation that fixes a compatibility
issue with the current version of iostat, which no longer contains a
colon after its table headers.
This patch is applied in order for the fix to be backportable to
existing stable releases. A final "proper" fix will be an upgrade to a
newer version of dd-agent, but that requires several other changes.
This fixes#40103.
