The tests timeout on AArch64 (e.g. [0] and [1]), likely because the QEMU
option "-vga virtio" isn't supported there (unfortunately I currently
lack access to an AArch64 system with NixOS to investigate).
This also affects the test for Cage but that one is already limited to
x86_64-linux.
[0]: https://hydra.nixos.org/build/144148809
[1]: https://hydra.nixos.org/build/144103034
(cherry picked from commit abb9ea73f7b2bd8d0862be47a0a1010d7644136b)
* The options tlsKey and tlsCert require being accessible by DynamicUser at runtime, which currently requires copying the files into the matrix service state directory. Fixing this might require breaking changes. Thus the module should not be included in a stable release.
Since 21.01, poppler supports PDF signing. As applications like okular
start to make use of that feature, nss support for poppler is enabled by
default to avoid unnecessary package duplication.
When building a `minimal` version of poppler, nss is disabled as well.
closes#120928
(cherry picked from commit 5f9862d524ee0ec1edcaa450fff4f54f4f5cc68e)
There's a bit of LaTeX math code in a comment in a test file:
```
.. math:: L = 10 \cdot \\log_{10}{\\left(\\frac{MS}{p_r^2} \\right)}
```
which of course should be `\\cdot`.
I could patch it, but I think skipping deprecation warnings in tests is
probably ok... (If nothing else it's easier.)
(cherry picked from commit dbc085cb44b90e4d533b6a24b7c9ca3c2c93d5e1)
There was an override of autoconfHook to use automake 1.15.x; I'm not
sure what changed since this package was added but it explicitly needs
1.16.1 or greater for the submodule libltdl.
Anyhow, just remove the override.
(cherry picked from commit 3e4aeea693d3473427d7a7a4b028c3a6a1a8908e)
The rpath structure for the bootstrap tools was reworked to minimize
the amount of rewriting required on unpack, but the test was not
updated to match the different structure.
Additionally [1] builds that use the bootstrap version of libc++
cannot find libc++abi if the reference includes the "lib"
component (ie, libc++ refers to libc++abi with
@rpath/lib/libc++abi.dylib).
[1] https://logs.nix.samueldr.com/nix-darwin/2021-05-18#4993282
Test failure observed on Hydra: https://hydra.nixos.org/build/143130126
(cherry picked from commit 38207735f425a06563482c726b25a0967bc9011d)
The default backend is chosen based on the content of the $DISPLAY
variable *and* a successfull call to libX11, loaded via dlopen().
The test fails because dlopen looks in /usr/lib and /lib, so matplotlib
falls back to a headless backend.
To reproduce try running:
$ nix-shell -I nixpkgs=$PWD -p \
'python3.withPackages (p: [ p.matplotlib ])' --run python
>>> import matplotlib.pyplot as plt
>>> assert plt.get_backend() == "TkAgg"
(cherry picked from commit 7e2ec8f8a15b77c14c185c4044e99ba05e8b6a38)
* Disable all tests of the NuSVC estimator that use memmap'd data
* build in serial on darwin
Resolves#121988
(cherry picked from commit cb2891b8c88705dfa5092cf992b7f64ae04da781)
