| Age | Commit message (Collapse) | Author |
|---|
|
Prevent locks caused by the zmq log handler reaching the zmq's HWM (High
water mark / buffer limit) and causing some components to block until
the buffer is empty (running the zmq handler for instance).
We run the zmq handler in the background all the time to prevent this.
- Resolves: #7222
|
|
Replace logging.getLogger with custom get_logger.
Remove some unneeded dependencies and reorder other.
|
|
- Related: #6359
|
|
If the quit() call is triggered by the system logout we should not
disable the autostart. Otherwise bitmask won't autostart on the next
session start.
- Resolves: #6424
|
|
FIX: Login attempt is made against previously selected provider.
Closes #6654.
Cherry-picked from release/0.8.x since it's a very annoying bug for the
testing cycles for the next release (0.9.0) -- kali.
|
|
|
|
|
|
|
|
FIX: Login attempt is made against previously selected provider.
Closes #6654.
Remove unused comments as well.
|
|
This causes (on certain scenarios) the app to segfault and it is really
hard to debug.
|
|
Now it's working and no longer segfaults, so we can enable safely this
feature again.
Stop tracking the gui connections on wizard.
|
|
|
|
This temporary hack prevents the app to have segfaults when the first
run wizard finishes.
For more information look at https://leap.se/code/issues/6597
|
|
Fix typo for signal disconnection.
The backend is the one who always creates the certificates. Either if it
is run separately or in a process in the same app as the frontend.
|
|
- factor out the signal tracking code as a helper class,
- move login logic from MainWindow to LoginWidget,
- add new signals to the LoginWidget to interact with MainWindow,
- add login sequence docs to LoginWidget class,
- improve docs for login methods,
- add LoginState class to handle login states,
- disable login button until data is entered,
- move some properties and actions to .ui file.
|
|
Closes #6150.
|
|
- cleanup some code,
- remove unused import,
- pep8 fixes
|
|
|
|
|
|
|
|
Hide the Advaneced Key Management menu from the ui file and comment out
the connection between the triggered action and the method that shows
the AKM window.
Closes #6087.
|
|
* Update transifex resource name.
* Use RightToLeft layout for Arabic language.
* Use better names and comments on i18n section.
* Use unicode to initialize widgets with text that otherwise (str) will
fail when Arabic (and most likely any other language with non-ascii
characters) is used.
Closes #5289, #6033.
|
|
|
|
|
|
Add a platform independent release_lock helper, so all the SO dependent
code goes inside the locks file.
Also, do some code cleanup.
|
|
* Emit the `eip_missing_helpers` signal when the user chooses "Don't ask
me again".
* We emit that signal when some helper file is missing, even if the user
doesn't want a warning.
* Do the update of the eip enabled status with some delay to give some
time the eip machine to start and do what it needs with the
buttons/labels and avoid the 'hide turn on button if missing files'
being overridden.
Closes #5945.
|
|
- Don't disconnect logout signals, so when the wizard is triggered the
UI can reflect the logout result.
- Restore the selected provider when 'other...' is selected *before* the
provider_changed signal is emitted to avoid that option to keep
selected in case of some error.
- Avoid the currentIndexChanged to be triggered when we load the providers.
|
|
Closes #5996, #5997.
|
|
Closes #5995.
|
|
|
|
|
|
|
|
Store the last 2 indexes on a deque and improve the 'select previous
provider' action.
|
|
|
|
|
|
'refs/remotes/meskio/bug/wait-for-eip-to-autologin' into develop
|
|
|
|
possible raise window when it is obscured. To hide, just close window.
|
|
|
|
eip machine was already moved into conductor
|
|
- Set the logger window parent,
- don't use an mainwindow instance variable to hold the window object.
This fix have the side offect that prevent multiple log windows being
created at the same time, but it does not causes any side effect or
problem.
|
|
This fix the bug that prevents pastebin to work.
Closes #5949.
|
|
Closes: #4035
This falls in the "quick" workaround category.
A proper state machine that extends the four basic connection states
is hence needed. We have to accomodate design to have a
connection-oriented state machine in the backend (I would favor a
twisted protocol for this), and a more lightweight one that conducts
the gui-level changes (ie, change the actions / buttons / labels
accordingly).
Since this "cancel" functionality has been long postponed, I chose
to do one more ugly hack here, that is, show and hide dance with a
button that just calls the bitmask-root to kill the vpn process. It
should work well enough until we get to the reorganization needed for a
clean process control for eip.
|
|
Use a simple sleep to wait until the services are stopped and then kill
the backend, always in a fixed time in order to avoid blocks/hangs on
the system session logout.
Closes #5870.
|
|
|
|
This fixes the issue where a system logout was blocked by bitmask since
we were interpreting the closeEvent as if the user clicked on the 'X'
button.
Closes #5870.
|
|
|
|
|
|
|
|
|
