And don't need to source the uevent files anymore either since $MAJOR
or $MINOR aren't used elsewhere.
[dezgeg: The reason these are no longer needed is that 0d27df280f
switched /tmp to a devtmpfs which automatically creates such device
nodes]
Use unpackCmd and installPhase instead of overwriting buildCommand.
This makes the derivation more readable, and adds binary stripping and
shebang patching of the output as a bonus :-)
Use makeWrapper instead of a custom wrapper script, and only export
environment variables that are used.
Set JAVA_HOME to ${jdk.home}, making it more generic and independent of
the specific jdk path being used (was a different one on zulu). This
fixes sqldeveloper on darwin, so add it to supported platforms
(plus the known-working x86_64-linux)
`git-annex` version 6.20180719 enhances the output of `git-annex info`
with a description and UUID of the repository which broke the testcase
`test_process_annex_info_batch` which parses the output of the info
command.
The testcase has been fixed accordingly and the patch was filed upstream
here: https://github.com/alpernebbi/git-annex-adapter/pull/6
I rechecked the functionality in a simple python environment with the
following expression:
```
with import ./. {};
python3.withPackages (ps: with ps; [ git-annex-adapter ])
```
Furthermore I tested the main functionality in a Python REPL within the environment.
The Darwin patch fixes tests that rely on `rev`, a binary which is
distributed in the `utillinux` package and not available on Nix with
Darwin.
To make the purpose of the patches for future contributors clear I
documented the purpose for each patch.
A few months ago I moved these patches to the new debian alsa instance [1], but
it looks like their `sha256`s on the tag at the remote have changed again.
It doesn't appear that debian's source remote is stable in the way we need it to
be; let's just vendor the patches to avoid future issues.
[1] https://github.com/NixOS/nixpkgs/pull/41769
