Currently, the failure to connect to an MTP device results in the UI displaying
an open device that appears empty. This change introduces a method
ConnectedDevice::ConnectAsync() that is expected to handle any connecting tasks
that could block asynchronously. Upon completion, this emits a ConnectFinished
signal that indicates success or failure. The row in the UI is only updated
after the successful response is received. Upon failure, DeviceManager will
clean up and the row in UI is left in the pre-connect state.
Currently, only the MtpDevice utilizes this mechanism. All other devices use a
default implementation that immediately reports success.
In 1.37.2, gvfs switched to URIs that remain consistent across USB device
re-enumerations. This removed the usb bus and device numbers from the URI. In
the case that these values aren't found in the URI, try to parse Unix device
name property and pass results as query params on the URL. Pay attention to
these params in MtpConnection.
See gvfs commits 3a7bb06b and efc76d0c for reference.
* Qt5::Test is not required in the global QT_LIBRARIES definition
* Qt5::DBus had already been optional, drop bogus pkgconfig search
This partially reverts commit 4321ecf7d2.
* Find X11 only once, in root CMakeLists.txt
Since we have HAVE_X11, use HAVE_X11 in cmake.
GError messages contain non-ascii characters. This normally just produces some garbage when we use the default QString contructor that assumes ASCII for logging. However, when a message includes the right double quote, UTF-8 sequence 0xE2 0x80 0x9D, the final byte is OSC. VT100 expects a command sequence to follow and stops echoing output until it sees ST or BEL character, which may never come. Thus, the console output is halted.
This change uses QString::fromLocal8Bit instead of depending on the default constructor. About half of the sites in the codebase had already been converted.
One side effect is that log messages are quoted. There are additional options to control this, but those were only introduced in Qt 5.4.
This tool was once used generate data/lyrics/ultimate_providers.xml from
the sites.js file. Judging by the file histories, recent changes has
been done directly in the xml file, bypassing this tool. Moreover,
sites.js has contained syntax errors for the past three years without
anyone noticing.
Furthermore, this tool is the only part of Clementine that uses
QtWebKit. Since support for this has been removed in both the master and
qt5 branch it won't build anyway.
GPodLoader and MtpLoader Error signals were connected to Error signals in their repective device classes, but the actual signal definition in ConnectedDevice was removed in a refactor several years ago. This change adds LoaderError slots to these device classes and reports the error in the manner of the refactored code.
Jonas Kvinge
aqua36
Clementine Buildbot
Dmitriy Bogdanov
John Maguire
Jim Broadus
Martin Delille
kz6fittycent
Andreas Sturmlechner
Rebecca Wallander
Mattias Andersson
Alex Kong