GitLab

When testing this with 2.2.31 it showed errors so better
not offer a button that only produces an error. According
to the GnuPG Maintainer this should work with 2.3

Having them with row stretch and aligned in the center had the
effect that the widgets moved when new recipents were added. So
we just stretch the second row now to ensure that it takes up
the available space.

When Kleopatra is started the readerstatus thread may check
for smartcards before the gpg-agent and scdaemon are started
which can take a while, especially on Windows. So we now
handle the GPG_ERR_ASS_CONNECT_FAILED and wait for it to
come up. GnuPG itself does something similar.

The accessibility improvements are finished for now and there
were several new features that should be distinguished.

This should fix the problem that the version number was not updated by
the release scripts. To avoid confusion also change the name of the
corresponding internal KLEOPATRA_VERSION_* variable.

# Conflicts:
#	CMakeLists.txt

Otherwise the Flatpak build complains

This reverts commit a4a84b9a.

I'm sorry, but this change breaks the tab order in the dialog. In fact,
it makes it impossible to reach any widgets except for the two dialog
buttons with the Tab key. I also don't see why passing a parent to the
constructor should be necessary. All widgets are added to a layout and
the layout takes care of setting the correct parent widget of the widgets.

* Add HtmlLabel which
  * sends a text selection event to the accessibility clients when
    a link is selected,
  * allows setting the color to use for links when rendering HTML
* Use HtmlLabel instead of QLabel for the decryption result overview
  and the decryption result details

GnuPG-bug-id: 5535

This condition is not mentioned in documentation or any examples for
QAccessibleInterface, it was apparently copied from qtbase's
qAccessibleFactory. That means it's most likely not necessary in plain
applications, otherwise it would be done by Qt itself.

We simply use the text color also for links. Links are still
recognizable by the underlining.

GnuPG-bug-id: 5535

Qt does not use the QPalette::Link color when rendering rich text.
Therefore, we do it directly via the style-attribute of the <a> tag.

GnuPG-bug-id: 5535

* Make ResultItemWidget a composite widget that can receive focus.
* Ensure correct focus order when adding a new result item widget.
* Add forceSetTabOrder helper for setting the focus order of widgets
  with any focus policy (including Qt::NoFocus).

GnuPG-bug-id: 5535

In particular, this ensures the correct focus order of the widgets.

GnuPG-bug-id: 5535

GnuPG-bug-id: 5535

Before, information on unknown recipients was shown (with early exit)
before the "No MDC" hint, i.e. in case of a symmetrically encrypted
message with additional public key encryption the "No MDC" hint
would not have been shown.

GnuPG-bug-id: 5535

In this case checking the compliance of the decryption (which didn't
happen) or showing the embedded filename (which is not available without
decryption) makes no sense.

GnuPG-bug-id: 5535

This change fixes a logic error/regression introduced with revision
54ae2682c871d24f16177e1b761de3fea46e59ce (in kdepim repo before the
split).

GnuPG-bug-id: 5535

Set useful tooltips (e.g. for screen readers) for the line edit in case
no certificates or multiple certificates match the user input.

GnuPG-bug-id: 5535

* Add UrlLabel, a label for displaying a single link, which
  * automatically gives focus to/selects the displayed link when the
    label gets focus,
  * sends a text selection event to the accessibility clients when
    a link is selected,
  * works around a bug in QLabel::focusNextPrevChild(false)
* Add AccessibleRichTextLabel which allows accessibility clients to
  retrieve the selected text of a UrlLabel
* Use UrlLabel for the Diagnostics/Show Audit Log label

GnuPG-bug-id: 5535

Check only enabled recipient widgets for valid key/group. List all
unresolved recipients in the error message.

GnuPG-bug-id: 5582

(cherry picked from commit c16459aa)

For some reason the screen reader "reads" the Close button even if it's
hidden. It seems that Qt somehow gives it (local) focus, but not
application wide focus.

GnuPG-bug-id: 5535

GnuPG-bug-id: 5535

GnuPG-bug-id: 5535

If one presses Return while a recipient input widget has focus, then the
wizard switches to the next page and the SignEncryptWidget emits an
operationChanged signal which triggers an update of the commit button.
This update needs to be ignored, so that the label text that is
cleared in initializePage() of the second page is not set again.

GnuPG-bug-id: 5535

This ensures that the VS-NfD compliance hint is read by screen readers
when the commit button gets focus or is read for some other reason.

GnuPG-bug-id: 5535

updateCommitButton implicitly triggers the creation of a custom wizard
button which inserts the wizard buttons into the focus chain before the
output widgets. To fix the tab order we need to call updateCommitButton
after the creation of the output widgets.

GnuPG-bug-id: 5535

If there are no secret keys in the keyring, then disable the "sign as"
option and the "encrypt to me" option. (The latter because we currently
require a secret key for the "encrypt to me" key.)
If there are no public keys in the keyring (and therefore also no secret
keys), then disable the "encrypt for others" option and check the
"encrypt with passphrase" option which is the only possible operation in
this case.

GnuPG-bug-id: 5535

If there are no keys in the keyring, then we check the visibility
of the requesters before the dialog is shown on screen, i.e. before
the dialog is visible, so that isVisible() returns false even for
those requesters that were not explicitly set to "not visible". What
we actually want to find is not the first visible widget, but the
first widget that is not explicitly hidden. Therefore, we need to check
for !isHidden() instead of isVisible().

GnuPG-bug-id: 5535