(cherry picked from commit 45153a99)

Apparently, the tags for these jobs changed from 'privileged' to 'asan',
so let's follow suit (so that our CI starts working again completely).

ci: Fix ASAN jobs

See merge request GNOME/gcr!59

Emilio Pozuelo Monfort authored Feb 05, 2017 and Niels De Graef committed Nov 07, 2020

The timeout length in the current test is too low and can make the test
fail if the system is too slow or too loaded. This test correctly stops
waiting when a result is received, so the timeout will not actually be
reached unless there is a problem.

[smcv: Expand commit message]
Fixes: https://gitlab.gnome.org/GNOME/gcr/issues/29
Bug-Debian: https://bugs.debian.org/846728


Signed-off-by: Simon McVittie <smcv@debian.org>

Increase test-gnupg-collection timeout

Closes #29

See merge request GNOME/gcr!23

The function is meant (by name and documentation) to check the slot info
flags, but instead it gets and checks the token tags.

Given that gck_slot_has_flags changed meaning in the previous commit,
add back a function that allows to get the slot token flags

Gck flags check fixes

See merge request GNOME/gcr!60

`GSimpleAsyncResult` is deprecated in favor of `GTask`.

That way, if we have an issue during one of these jobs, we can check for
example the meson logs to see what went wrong.

Get rid of most GSimpleAsyncResult uages

See merge request GNOME/gcr!29

This allows applications to set mnemonics, which is needed for
accessibility purposes.

https://gitlab.gnome.org/GNOME/seahorse/-/issues/296

ui: Set "use-underline" for GcrImportButton

See merge request GNOME/gcr!63

This avoids circular dependencies, such as gcr -> gpg2 -> pinentry -> gcr

This allows us to lose the hard dependency on `gtk-x11`.

GDK provides an interface to the _xdg-foreign_ protocol extension, which
exactly allows to export a handle that another window can use to set
itself transient to.

Also bump the minmal GTK version to 3.22, since that is the version that
adds the necessary methods to do this in GDK.

`frob-prompt` and `frob-system-prompt` aren't used in practice, but they
give a good idea on how to export a handle when dealing with Wayland
windows.

Replace it with the dbus-daemon package and just use `dbus-run-session`
to run the tests in a separate session bus.

It's possible that a GTask that has been called with
`g_task_run_in_thread()` still has a reference held by that thread after
the `_finish()` function has been called. This is a little problematic
for tests that check for the final reference, so use this as a
workaround instead.

See https://gitlab.gnome.org/GNOME/glib/-/issues/1346 for more info

Create a workaround for tests and GTask

See merge request GNOME/gcr!65

`g_assert()` can be (accidentally) disabled if someone uses
`G_DISABLE_ASSERT` when compiling gcr, so that renders most tests
useless. As a nice side effect, we get better error report when
something goes wrong in a test.

gck: Don't use g_assert() in tests

See merge request GNOME/gcr!64

It's been deprecated for some time now and just gives us an extra load
of compiler warnings, so just define a little `MockProperty` struct with
the same layout.

Apart from being less code, this actually gives us a nice performance
improvement. Up until a few years ago, if you pass `NULL` as the
marshaller for a signal, GLib would fall back to
`g_cclosure_marshal_generic` which uses libffi to pack/unpack its
arguments. One could avoid this by specifying a more specific
marshaller which would then be used to immediately pack and unpack into
GValues with the correct type.

Lately however, as a way of optimizing signal emission (which can be
quite expensive), GLib added a possibility to set a va_marshaller, which
skips the unnecessary GValue packing and unpacking and just uses a
valist variant.

Since the performance difference is big enough, if the marshaller
argument is NULL, `g_signal_new()` will now check for the simple
marshallers (return type NONE and a single argument) and set both the
generic and the valist marshaller. In other words, less code for us with
bigger optimizations.

In case you also want va_marshallers for more complex signals, you can
use `g_signal_set_va_marshaller()`.

We have a modern, saner build system in place with Meson, so let's drop
autotools before releasing the next version of gcr.

This is more in line with other GNOME projects.

Remove Autotools

See merge request GNOME/gcr!58

It's been discontinued for almost a decade now, and we no longer need to
ship the file since we removed our autotools build.

This makes it show up in the GitLab UI, and is a bit more user-friendly
as the markdown will be rendered.

GckCall uses its own implementation of threads pool to handle the async
calls, now that we've GTask this code can be simplified by reusing
GLib code.

I didn't want to change the API in this commit, even if private not to
mix changes together, so the functions still are used as they used to be

The main difference is that the async_pre and async_ready prepare a
GTask instance, while the async_go starts a thread using the gtask
function.

Callback functions needed to be adapted to the new GAsyncResult type.

Tests needed also some tuning as the underneath task holding a reference
to the source object might be finalized at later stage, when the
thread-related data is removed, as per this we may wait a bit to check
whether a source object gets finalized, this is due to GNOME/GLib#1346.

This is to mimic GTask more and so we can avoid using a temporary task
only for temporary storing an object pointer.

Never return or expect arguments as paramter for gck_call_* functions,
instead use gck_call_get_arguments to interact with them when needed

GckCall: Simplify the code by using GTask based implementation

See merge request GNOME/gcr!61

(cherry picked from commit 4cafb79cb2b4a7f8d28ac74df732355dfb0ccaba)

gcr/key-mechanism: Port to GTask

See merge request GNOME/gcr!66

Seahorse is able to handle hkps keyservers.

FEATURE: add hkps://keys.openpgp.org to keyserver defaults

See merge request GNOME/gcr!62