Fixes#1926 : When an IPC event was processed locally, if the window was closed before the core could start, the event handler would be forever stuck in the background waiting for the core to start. We fix this by substituting QApplication::quit() by a Nexus::quit() function and a Nexus::isRunning() function, which gives us a condition for exiting blocking processEvents() loops. We cannot simply use QApplication::quit(), because this function has no effect before the start of the event loop.
The problem was further exacerbated by the Tox URI event handler being (incorrectly) blocking. The IPC owner would block in this event handler, and the sender of the event would give up waiting and process the event itself a second time, potentially triggering the first bug. We fix the event handlers accordingly to be (mostly) non-blocking.
Also fixes a related deadlock between ~Core and ~Profile in the case of an early exit
This commit removes historykeeper and code only used by it.
BREAKING CHANGE:
qTox can't open old historykeeper chatlogs after this commit. Use
previous versions to migrate to the new database.
Revert needed, since otherwise there is no way to do automatic sorting
of includes.
Also reverted change to the docs, as leaving it would make incorrect
docs.
In case of conflicts, includes were sorted according to the coding
standards from #3839.
This reverts commit b4a9f04f92.
This reverts commit 5921122960.
There is a reason why magic length is not exposed via TES header –
knowing it is not needed. Client should check whether file is a TES with
a define that is provided in the public API.
Fixes an issue where qtox refused to load profiles while using
the -p switch on the command line. This happened in cases where
there was no .ini file for the specified profile present.
qTox will automatically import the old history on startup.
This new database code is much more robust.
It is very resilient and will not corrupt or disappear after a crash or
power failure, unlike the old code.
The on-disk database format is also much more compact now.
The database sync option in the advanced settings has been removed,
we know run many database operations asynchronously so performance
should not be a problem anymore, but we always ensure resiliency
in case of abrupt termination, so there is no tradeoff anymore.
And make it saner by not having one global password that has to be set before encrypting/decrypting, which is as racy and poorly designed as it gets
Fixes#1917 's immediate symtoms, which some potential for other regressions due to the mess that is encrypted persistence currently
We now use a binary serialized format to save space and allow clean encryption of the user settings.
All the settings can (and should) be edited from the GUI so there is no loss of functionnality.
It can still read the old .ini format, and will seamlessly upgrade to the new format.
Fixes#1810
