With Qt 5.6, build fails with error below.
This is because in Qt 5.6, the 3rd argument is not optional. Starting from Qt
5.7 the default value for the 3rd argument is nullptr, so setting it to
nullptr.
https://doc.qt.io/archives/qt-5.6/qaction.html#QAction-2https://doc.qt.io/archives/qt-5.7/qaction.html#QAction-2
Error:
src/gui/tag/TagView.cpp:79:38: error: no matching constructor for initialization of 'QAction'
auto action = menu.exec({new QAction(icons()->icon("trash"), tr("Remove Search"))}, mapToGlobal(pos));
^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* Fix detecting AES-256/GCM cipher, fixes#8964
When you generate a ssh key using the aes-256/gcm cipher, the cipher name in the keyfile includes an @openssh.com at the end.
* Use separate iv length for getting iv data, the assumption that the block size and iv size are equal does not hold for every cipher mode (e.g., GCM)
* Disable AES-256/GCM for now in ssh keys
Currently, the granularity for the botan gcm implementation is too large. To fix a problem with another algorithm in the library, they are multiplying
the blocksize, so by default the granularity is 64. This causes issues since the encrypted data in the key is only guaranteed to have a length that is a multiple of the block size (16).
currently, when keepassxc is not running, the command `keepassxc --lock` opens a new keepass window and blocks until the window is closed.
Especially in locking scripts this is rather unexpected and Ican't think of a case where someone explicitly starts keepass with --lock and wants this behaviour.
Rather --lock should always ensure, that there are no unlocked instances and exiting afterwards
* Fixes#8626
* Also remove old feature to set the title of a new entry to the current search text. This only made sense before advanced searching was made available.
* The original method dumped the changelog lines into the command which overflowed the command line length. This method uses a temporary file for the tag message.
This was noted as a problem in several issues and it finally occurred to me and I traced it to the fact that a timing issue sometimes allowed the file watcher to trigger a "file changed" alert right when saving starts. I fixed this by moving where the mutex lock is made for saving and preventing database reload during a save operation.
