avoid pinentry failures when bound to libsecret and the secret service is not running.  Thanks, Stef Walter.

Upstream version 0.9.7

--
Suggested-by: Daniel Kahn Gillmor

* secmem/util.c (lower_privs, raise_privs): Remove commented
functions.

* tty/pinentry-tty.c (password): Remove trailing space.

* tty/pinentry-tty.c (confirm): Flush the output after printing the
prompt.  Echo what the user typed.  If the user's selection is
invalid, indicate as much.

--
Signed-off-by: Neal H. Walfield <neal@g10code.com>

* qt/main.cpp (qt_cmd_handler): Keep utf8 pin byte array.

--
The pin pointer was invalid after the UTF8 conversion
as the converted byte array only was temporary in that
call and the data pointer becomes invalid after it's
destruction.

Commit message amended by Andre Heinecke. The original
commit message was:

const char* does not keep the temporary string returned
from toUtf8() active, since it doesn't keep the reference counter
set. So you usually just copy out garbage instead of the
pin that was entered. Just keep the QByteArray which keeps
the reference active.

GnuPG-bug-id: 2133

use maintainer scripts to leave pinentry-qt4 in place, even though it
only symlinks to pinentry-qt.  Purging pnentry-qt4 will get rid of it
from /etc/alternatives too.

* m4/qt.m4 (FIND_QT): Remove spaces in variable assignment.

--
GnuPG-bug-id: 2105

* m4/qt.m4 (FIND_QT): Quote initialization of have_qt5_libs var.

--

Problem reported by Kristian F.

GnuPG-bug-id: 2105

we're also providing a transitional package for the old pinentry-qt4
name -- upstream is clearly intending to switch to the new
pinentry-qt.

I've reduced the default priority for pinentry-qt in
update-alternatives from 95 to 80, since upstream seems to be focusing
on GNOME integration with a higher priority than KDE at the moment.

* m4/qt.m4 (FIND_QT): Add --disable-pinentry-qt5 option to
disable qt5 support even if it is available.

--
As requested by Kristan F.

GnuPG-bug-id: 2105

* pinentry/pinentry.h (struct pinentry): Add field invisible_char.
* pinentry/pinentry.c (pinentry_reset): Restore that.
(option_handler): Add option "invisible-char".
* gtk+-2/pinentry-gtk-2.c (create_window): Set the invisible char if
given.
--

The default GTK+ invisible character is pretty wide so that only ~16
characters show up in the smalles pinentry.  This is a bit annoying if
the passphrase is a longer due to the missing visual feedback.  By
using #x2022 ("•") this allows for ~23 characters and a plain "*" even
3 more characters.  Tastes are different so we allow to change that
character at runtime.

Signed-off-by: Werner Koch <wk@gnupg.org>

Upstream version 0.9.6

--

Signed-off-by: Werner Koch <wk@gnupg.org>

* tty/pinentry-tty.c (password): Print any supplied error message.

* tty/pinentry-tty.c (dump_error_text): New function to display error
messages.
(confirm): Use it.
(password): Likewise.

* pinentry/pinentry.h (pinetry_loop2): Clarify comment.

* qt/main.cpp: Import static platform plugins when necessary.

--
As pkg-config does not expose all libraries needed to link
qt5 statically with this patch it is possible to get a
static build by setting the LIBS variable correctly.

* m4/libassuan.m4: Respect SYSROOT Variable.

--
This change makes the libassuan lookup work similar
to the libgpg-error lookup simplifying cross-compiling
a bit.

* pinentry/pinentry.c (pinentry_loop2): Use assuan_fdopen for pipe fds.
* pinentry/pinentry.h (pinentry_loop2): Mention this in the comment.

--
For Windows calling assuan_fdopen is neccessary as this does
some internal platform specific stuff (get_osfilehandle).

This issue was introduced by using the real libassuan instead
of the built in variant.

* tty/pinentry-tty.c (password): If the user repeated the passphrase
and they matched, then set PINENTRY->REPEAT_OKAY.

* tty/pinentry-tty.c (read_password): Fix buffer resizing bug.