Update versions & CI image for 47 development branch

See merge request !253

Part-of: <!253>

Part-of: <!253>

https://www.youtube.com/watch?v=mYKLvYGqaC0

copy-worker: Run later, but before gnome-keyring-daemon

See merge request !250

In gnome-build-meta#821 we saw that running so early, with
DefaultDependencies=no, causes issues when GVFS (via GFile) tries to use
D-Bus.

!249 configured GIO to use only the local VFS,
but still, there is no need to run quite so early: the key thing is to
copy the user's keyring before the keyring daemon.

Adjust the unit to attach to default.target (not basic.target),
reinstate default dependencies, and explicitly run Before
gnome-keyring-daemon.service.

See !249 for further discussion.

Part-of: <!250>

contributing: Add instructions on how to build g-i-s

Closes #170

See merge request !238

- Guide people more pointedly towards GNOME Builder.

- Remove instructions on how to use toolbox: the toolbox documentation
  covers that.

- Adjust instructions about showing the welcome page.

- Combine with UNDER_JHBUILD tip, added in commit
  83ffabff.

Part-of: <!238>

𝅳Automeris naranja authored Feb 14, 2024 and Will Thompson committed Apr 22, 2024

Closes #170

Part-of: <!238>

Avoid a deadlock starting user session

Closes gnome-build-meta#821

See merge request !249

In !245 the
gnome-initial-setup-copy-worker.service moved to before `basic.target`.

We now need to ensure it doesn't try to use the D-Bus session bus
which is not ready until after `basic.target` completes.

Be more consistent about setting state of 'onwards' buttons (#199)

Closes #199

See merge request !247

CONTRIBUTING: Mention the env var for "Mock mode"

See merge request !248

This code was added in 69914dde

 but is not currently used. It is
intended to be used in future by pages which manage their own
'onwards' button. It seems incomplete - it only changes the
state of one of the three possible header buttons, and only its
visibility, not its sensitivity. It seems safer to ensure all
three possible header buttons are not visible and not sensitive.
We can conveniently do this by sending `set_navigation_button`
a widget which is none of the three buttons.

Signed-off-by: Adam Williamson <awilliam@redhat.com>

Make `update_navigation_buttons`'s behaviour more consistent and
clearer by always explicitly setting both the visibility and
sensitivity of all three possible buttons (forward, accept, and
skip).

In the current code, the logic is:

* If the page is complete, set the next widget sensitive and
visible; set the other two buttons not visible, but do not change
their sensitivity
* Otherwise, if the page is skippable, set the skip button visible
and the other two buttons not visible, do not change any button's
sensitivity
* Otherwise, set the next widget visible but not sensitive; set
the other two buttons not visible, but do not change their
sensitivity

This seems pretty inconsistent, and causes a clear bug, #199.
@lnicola proposes !246 to fix #199

. It changes the second
behaviour in the list to:

* Otherwise, if the page is skippable, set the skip button
visible and the other two buttons not visible, and set the next
widget as not sensitive

which addresses the immediate bug, but overall, I still feel like
the logic here is pretty inconsistent. To me it feels clearer and
overall safer to always touch both the visibility and sensitivity
of all three buttons, in a way that is (hopefully) consistent
with their intended state:

* If the page is complete, set the next widget visible and
sensitive; set both other buttons not visible and not sensitive
* Otherwise, if the page is skippable, set the skip button
visible and sensitive; set both other buttons not visible and
not sensitive
* Otherwise, set the next widget visible but not sensitive; set
both other buttons not visible and not sensitive

Signed-off-by: Adam Williamson <awilliam@redhat.com>

copy-worker: Don't try to copy GOA config file

See merge request !244

copy-worker: Ensure worker has completed before consumers

See merge request !245

Services such as gnome-keyring-daemon that consume the files may be
systemd socket or D-Bus activated. In order to ensure the generated
config files are in place before any consumers, run the worker before
basic.target completes. While here, sprinkle in some debugging messages.

Helps: gnome-keyring#137

Online Accounts support has been removed in this cycle, so this file is
not expected to exist. Trying to move it when it doesn't exist is
harmless, but unnecessary.