That way sending mail with mutt works on NixOS too, without having to
manually change mutt's config on your system. sendmail will be found in
$PATH at runtime, and thus the setuid wrapper will be used, if
available.
Upstream changes since 0.8.2:
* Performance improvements for collection browsing.
* Improved memory footprint.
* Show cloud collections above your friends in the sidebar.
* Keep looking for alternative sources even when we found a perfect match.
* Fixed crash in network code.
* Fixed repeat one/all icons.
* Be more lenient about accepting JSPF and M3U playlists.
* Improved support for compilation albums.
* (Windows) Fixed support for Windows XP.
* (OS X) Fixed settings dialog appearance on Yosemite.
* (OS X) Fixed SSL errors on Yosemite.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
* mediainfo{-gui} 0.7.72 -> 0.7.73
* libmediainfo 0.7.72 -> 0.7.73
* libzen 0.4.30 -> 0.4.31
From the mediainfo ChangeLog:
```
Version 0.7.73, 2015-04-09
+ BPG: basic support, thanks to Kurtnoise
+ CAF: basic support of Apple Core Audio Format, sponsored by FlavorSys
+ JPEG-2000: Display of profile (Rsiz)
+ JPEG-2000: detection of XYZ colorspace (based on D-Cinema profile)
+ FFV1 in MOV: more details (version...)
+ MOV/MPEG-4: handling of clcn (little endian nclc) Color parameter type
+ #P84, Matroska: Add TEXTST support to the MKV Parser, thanks to Kurtnoise
+ #P85, MPEG-TS: Add TEXTST support to the MPEG-PSI Parser, thanks to Kurtnoise
+ MediaInfoDLL interface: clean up, more debug features added
+ MediaInfoDLL interface: Giant threads lock removed
+ #F460, VC-3/DNxHD: detection of RGB 444 and other SMPTE ST 2019 (2014) new CIDs
+ VC-3/DNxHD: version number (HVN)
+ Clean aperture size is move from width/height to its own field
+ HEVC: tier
+ MXF: writing library and writing application fields are cleaned up
+ ProRes: support of 4444 profiles
+ CAP: detection of CAP files from Cheetah and Lambda
x B886, XML and HTML outputs were broken
x B902: EBU AspectRatio invalid in case of non classic numerator:denominator format
x #B758, VC-3/DNxHD: wrong color space and subsampling in case of unknown CID, now empty if format version is not known
x #B903, MXG: Incorrect timecode track used for AS-11 DPP MXF files
x #B904, MXF: Handling repetition of Header Metadata in MXF files
x MXF: AFD value was wrong (displaying the complete byte, but AFD is only 4 bits of this byte)
x DTS: some streams in Little endian were not detected
x MPEG-4: some files were having an incorrect frame count
x AVC: Some SCTE 128 caption streams were displayed twice
x BMP; accepting files with file size information set to -1
x RF64: samplesCount was not always right
x MOV: avoid wrong parsing in case of "colr" atom with "prof" color parameter type
x DCP/IMF: ID has now the AM/PKL CPL order in order in all cases (not only when there is more than one CPL)
x #B893, MXF: Crash with Panasonic P2 3.0 files
x DPX: time information was with native ":" character
x Images sequence: "Delay" field is filled from the number in the name of the first file
x FLV: some files were not having the right duration
x DPX: Cineon files were detected as DPX Version 1, version number was not corresponding to the real version
```
Primecoin is an altcoin which uses prime-searching as its proof-of-work.
It is the first energy-multiuse altcoin - normally, the proof-of-work algorithms
in altcoins are useful only for the coin itself, but the Primecoin algorithm
is very useful to mathematical research..
Release notes: http://slic3r.org/releases/1.2.6
Slic3r needs additional dependency, 'threads'. Add it.
In addition to bug fixes and some new features, this update stops Slic3r
from printing this on startup:
Running Slic3r under Perl >= 5.16 is not supported nor recommended
We don't have perl < 5.16 anymore, so we better update slic3r.
Slic3r dropped this dependency in v1.1.0 and we have v1.2.x.
And more importantly, BoostGeometryUtils is currently broken (fails on
newer Perl versions), so this patch unbreaks Slic3r.
This was caused by multiple things: First, the module-path was wrong in
the release. Second, when modules tried to load stumpwm, asdf searched
for its sources in /tmp/nix-build-*.
Both of these issues are fixed by a nix-specific patch that tells adsf
to *never* try to load stumpwm (and others) from the filesystem. This is
fine as those modules are already available in the image anyway.
We also refactor some stuff & clean up the build. Stumpish works now
too.
Upstream changelog:
* GUI: in the snapshots pane, protect the age of snapshots against
wrong host time
* NAT Network: fixed a bug which prevented to propagate any DNS name
server / domain / search string information to the NAT
network (4.3.24 regression)
* NAT Network: don't delay the shutdown of VBoxSVC on Windows hosts
* Mouse support: the mouse could not be moved under rare conditions if
no Guest Additions are installed (4.3.24 regression)
* Storage: if the guest ejects a virtual CD/DVD medium, make the change
permanent
* VGA: made saving secondary screen sizes possible in X11 guests
* SDK: fixed the VirtualBox.tlb file (4.3.20 regression)
* rdesktop-vrdp: make it work with USB devices again (4.3.14
regression)
* USB: fixed a possible BSOD on Windows hosts under rare conditions
* iPXE: enable the HTTP download protocol on non-Linux hosts
* Mac OS X hosts: don't panic on hosts with activated SMAP (Broadwell
and later)
* Linux hosts: don't crash Linux 4.0 hosts
The same with bug IDs can be found at:
https://www.virtualbox.org/wiki/Changelog
Tested on my machine using the virtualbox NixOS VM test.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This is espacially cruicial when it comes to Nix 1.9, where we even have
a more restrictive /nix/store. In any event, VirtualBox in hardenend
mode doesn't have to check the /nix/store path, because it's read-only
on NixOS systems. So this check would not introduce more security but
more hurdles, thus I'm removing it (of course _only_ for /nix/store).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This is a web browser plugin that allows one to use their Garmin device
on the myGarmin website. It provides a Linux version of the Garmin
Communicator plugin.
The autoconf build system for poppler does not support building the
wrappers separately, so this slightly enlarges the size of closures. To
compensate, the command-line utilities have been separated into their
own package.
Looks like this is the only way to do what 6258728318
was expected to do that doesn't break things and doesn't look like a complete hack.
The problem is that different `fetch`ers generate very different results (from
archive files to plain source trees) and allowing overriding `src` in top-level
derivation when there is a derived python package is no trivial matter.
This also removes qca2_ossl, because it's now bundled with the main QCA
package and we can now drop all those patches and build system fixes,
because they have switched from autotools to cmake.
Tested against a few builds like PSI and Tomahawk.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Also requires enabling iana-etc. I'm shooting for "unix" platforms which
seems reasonable. Not sure why this was restricted to linux originally --
the history doesn't tell.
