lp:ubuntu/trusty-proposed/kubuntu-driver-manager
- Get this branch:
- bzr branch lp:ubuntu/trusty-proposed/kubuntu-driver-manager
Branch merges
Related source package recipes
Branch information
Recent revisions
- 10. By Harald Sitter
-
* Fix debug output typo s/beause/because/
* Implement QApt transaction prompt signals
+ This adds three new qslots to drivermanager:
- provideMedium
- untrustedPrompt
- configFileConflict
+ All three are copied from libmuon's ApplicationBackend and
adjusted in the least invasive manner
+ All three trigger a KMessageBox to query the user for a decision on
the respective apt/dpkg-level issues
+ The KMessageBoxes anchor to qApp::activeWindow as DriverManager is
supposed to not contain UI logic and as such has no explicit method
to get the master window id
+ These slots add new strings that were previously not available for
translations, but should eventually get picked up by launchpad.
+ Module.cpp manually adds 'libmuon' as a translation catalog to draw
translations from, while we have no translations of our own. This is
to prevent the newly introduced strings from going untranslated.
This relationship is not refelected on a packaging level since in due
time our own language-packages should contain translation, plus the
prompts are incredibly hard to trigger so even in an untranslated state
99.9% of users will not ever see them. If we can get the translations
from libmuon it's nice, if not we'll likely not care given the impact.
LP: #1315670 - 9. By Rohan Garg
-
[ Harald Sitter ]
* DriverWidget: do not list builtin drivers as not selected.
+ This avoids breaking the system by trying to remove essential packages
such as xserver-xorg-video- nouveau for being not selected.
LP: #1311583[ Rohan Garg ]
* Add a apport hook to provide more debug info when reporting bugs - 8. By Harald Sitter
-
* Prevent crashes by properly locking the GUI while applying changes
LP: #1311075
+ New function DriverWidget::setSelectionEn abled, disabling or enabling
all QAbstractButtons inside the respective DriverWidget, leaving the
header label unaffected to give a nicer visual experience.
* Prevent state confusion caused by allowing 'defaults' to trigger a reload
even when the manager is currently busy applying changes. Since we have
no button control inside the KCM we simply turn load() noop while the
manager is active. This is unfortunately not the nicest behavior, however
the module will reload whenever the manager finishes, so implicitly we
will simply execute the requested action once the manager is idle again.
LP: #1311079
+ Introduce new function DriverManager::isActive evaluating the transaction
pointer (nullptr when idle). - 6. By Harald Sitter
-
[ Rohan Garg ]
* Use a checkbox when there's only one driver[ Harald Sitter ]
* Set dbusinterface timeout to highest possible value to prevent timeouts
when the dbus helper is tediously slow (LP: #1303209)
Looking back that was probably why we opted for an async connection earlier - 5. By Rohan Garg
-
[ Harald Sitter ]
* Fix dbus helper timing (hopefully)
+ Don't query drivers-common in ctor for no reason (block startup and
everything).
+ Force the connection from the timeout timer to qapp::quit to be queued
this gives us at least one additional loop run until it is actually
called (should not make much difference).
+ Suspending and restart of the timeout timer is now done via a finally
statement, this is to ensure that return actually has finished by the
time we restart the timer as large chunks of data can take a while to
throw at the daemon etc.
= These changes combined should prevent the dbus helper from timing out
before the data has arrived or is sent. at the very least it makes a
system that previously had timeouts work even with a 5s timeout (i.e.
half of what we use in production). - 4. By Harald Sitter
-
* Depend on python3-qt4 for the auto-quit logic in the helper LP: #1293948
* Refactor
- dbus helper now only has one function
- dbus interaction uses explicit type demarshalling
- new struct types Device and Driver representing the explicit dbus types
- new class DriverManager encorporating most logic bits of Module such as
dbus and qapt interaction
- DriverWidget QApt queries moved to DriverManager's dbus handling, but
continues to query qapt packages directly for name and such nonesense
- Manual driver listing is no longer explicitly dependent on package
findability
- DriverWidget changes detection now happen through iterative queries
in Module::possiblyChange d, fixing change tracking
- Module continues to contain Debconf handling, pipe id merely passed to
DriverManager
- New Module functions enableUi and disableUi containing reoccuring logic
- DriverWidget does not implicitly create left-to-right strings anymore
- Module::load is now the grand unified way to trigger a (re)load of data
while DriverManager::refresh synchronizes the reload requests with
QApt/Xapian init, fixing all likely and less likely init order problems
- New struct Device provides deep qdebug stream operators to debug
stuff conveniently
- Module includes have been cleaned up slightly
- Drivers are now not only installed, but also removed, this effectively
enables actual driver change as the way management works is that the
consideration is that only one driver is installed at a time (except for
the free version), which also is why manual always wins active
detection as it simply prevents any other driver from being considered. - 3. By Rohan Garg
-
* New upstream snapshot
- Bug fixing around
+ Better strings LP: #1282095
+ Update xapian cache before opening LP: #1282087
+ Fix memory leak
+ Drop useless keepAlive function LP: #1282092
+ Timer should start/stop before each function so that the script
doesn't quite mid way LP: #1282090
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)
