If there was no older file than $NIX_BUILD_TOP this would result in a
warning, e.g. with nix-info.
```
/nix/store/15kgcm8hnd99p7plqzx7p4lcr2jni4df-set-source-date-epoch-to-latest.sh: line 13: [: : integer expression expected
```
I've introduced the plugin and have been maintaining it ever since, so
it's time to make myself the official maintainer in order to avoid
confusion about who to address when issues about the alternatives plugin
arise.
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @wisp3rwind
This introduces the following upstream changes:
* The package is now on PyPI
* Require at least beets v1.4.7
* Update album art in alternatives when it changes
* Python 3 support (Python 2.7 continues to be supported)
* Support the format aliases defined by the convert plugin ('wma' and
'vorbis' with current beets)
* Bugfix: Explicitly write tags after encoding instead of relying on
the encoder to do so
* Bugfix: If the formats config option is modified, don't move files
if the extension would change, but re-encode
I updated this because I was pinged by @wisp3rwind about moving back to
@geigerzaehler's repository at [1].
This is what @wisp3rwind wrote in the comment[2] (which was originally
directed to @Profpatsch):
(I hope you're the one to bug, or at least can ping someone else), I
just noticed that you switched the NixOS package to my repository.
Would you please switch it back to this repo soon-ish? The code here
is better tested, and [3] is handled less elegantly on my fork since
it requires changes to the configuration. The latter are undocumented,
but whoever has bothered to take a look at the code might end up with
(harmless) unused config entries.
So in essence we're now back to the original upstream repository again,
which I changed to @wisp3rwind's fork in 29e89248bf
because it fixed issues with Python 3.
Stripping the long_description from setup.py also doesn't seem to be
required anymore, but I didn't investigate why (might be because either
our Python tooling now sets a default language or the README simply no
longer has non-ASCII characters).
[1]: https://github.com/geigerzaehler/beets-alternatives
[2]: https://github.com/geigerzaehler/beets-alternatives/issues/23
[3]: https://github.com/geigerzaehler/beets-alternatives/pull/27
Signed-off-by: aszlig <aszlig@nix.build>
Since 0f38d9669f, the default Python
version for Python 3 is now Python 3.7.
It has been a while since beets had a new release, but the fix for
Python 3.7 is already in master (and it's also rather small), so I
decided to cherry-pick the commit as a patch.
I've built the package along with its tests and they failed at first,
but the errors were unrelated. So I disabled the tests for pylint, as
they're failing right now.
In addition I also needed to temporarily revert
0d2f06ae3a, which supposedly should fix
issues with Python 2 but aparently breaks Python 3 support and during
the beets tests we get a ModuleNotFoundError for the "_gi_gst" module.
However I didn't further investigate why this happens, as I'm time
constrained right now. But after disabling the pylint tests and the
revert of the mentioned gst-python commit, the beets tests succeed.
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @jtojnar, @lopsided98 (for introducing the gst-python change)
Cc: @domenkozar, @pjones (other beets maintainers)
Although the tests are passing locally, it seems as the excessive
filesystem usage causes several build failures and timeouts in the Hydra
and OfBorg infrastructure:
* https://hydra.nixos.org/build/84374861 (python3 on linux.x86_64)
* https://hydra.nixos.org/build/84368459 (python2 on linux.x86_64)
Some of these tests are failing after several seconds though, but I
couldn't identify a pattern and I'm not overly surprised that a FTP
library has impure tests. However the API seems to be usable in a Python
{2,3} environment, so it should be safe to use even with disabled tests.
```
b Batch mode. Optimized for huge recursive copies, but less secure if a crash happens during the copy.
```
It seems the "less secure if a crash happens" does not need a crash to
happen.
With batch mode:
```
/[...]/.
Start (0) does not point to parent (___)
```
For pretty much everything copied in.
Without batch mode, everything passes `fsck`.
See #51150
Some trivial builders use the name attr to choose the exec name
produced. For example nixos-install,
nixos-install = makeProg {
name = "nixos-install";
src = ./nixos-install.sh;
nix = config.nix.package.out;
path = makeBinPath [ nixos-enter ];
};
When cross compiling, this puts the prog in,
/bin/nixos-install-powerpc64le-unknown-linux-gnu
When separateDebugInfo = true & !stdenv.hostPlatform.isLinux, we
always gave an error. There is no reason to do this. Instead, just
don’t add separate debug info when we aren’t on Linux.
