GitLab

Reviewed-by: hjk

Reviewed-by: hjk

Split for testing/scripting purposes.

missing

doing it in 1.3 as well to avoid possible later conflicts

Fixing various problems with breakpoints and opened files caused by CDB
returning lower-case file names.
Task-number: QTCREATORBUG-438
Reviewed-by: hjk <qtc-committer@nokia.com>
Acked-by: mariusSO <qt-info@nokia.com>

Make it possible to trigger a debug-attach in a running instance
of Qt Creator by means of -client, use that in qtcdebugger, thus
enabling it to attaching to crashing executables in run mode
(Windows/CDB).

Modify IPlugin and application so that the complete command line is
serialized and transmitted to a running instance via QtSingleApplication
if -client is specified. Introduce IPlugin::remoteArgument and use that
in core for opening files and in the debugger for attaching.

Use -client in qtcdebugger with some logic to keep it alive as long as
the debuggee, make CDB break in that case as the events are not replayed
correctly in case the debugger is not spawned by the registered handler.
Rubber-stamped-by: con <qtc-committer@nokia.com>

Reviewed-by: hjk
Task-number: QTCREATORBUG-399

Check for correct state (InferiorStopped).

Use debugger's 'gu' command instead of temporary breakpoint as a
workaround until it is properly supported by the
SetExecutionStatus() API.

Reviewed-by: hjk
Task-number: QTCREATORBUG-240

this includes:
- move the gdb ownership back to the engine (thus strip down the
  adaptors as far as possible)
- make gdb startup synchronous
- make adapter shutdown synchronous
- fix the state transitions relating to shutdown

As otherwise breaking when attaching to running processes
does not work as the startup trap is not received.

... with an event loop with disabled user input to keep
the GUI alive.

after refactoring the trk adapter startup, the split was not necessary
any more. in fact, it seemed fairly arbitrary at this point.

- Turn off initial breakpoint when attaching (terminal) and
  additionally ignore it before the first startup complete
- Refuse to step  the artifical thread created when
  interrupting
- If possible, switch to thread 0 when interupting (do not
  use artificial thread)
- Ignore non-fatal exceptions in dumper calls (stray Startup-
  complete-traps)
- Use right thread for dumper call loading
- Rename windows exception code enumeration
- Turn off modification off watch-data children by setType,
  show empty lists correctly.
- Verbose warnings about inserting invalid watch data.

Add utility to figure out DLL/Product versions.

Use new call syntax of 6.11. for dumper call loading. Execute
Dumpers in a single thread (current) if at all possible (not
in some WaitFor or artificial break thread). Show one more
frame in threads view if it is FastCallReturn. Fix step into
(ignore one event), log some more exceptions. Generally log
exceptions to the debugger windows to be able to see stuff
like DLL missing, etc.

Also move Designer::Internal::FormWindowEditor ->
Designer::FormWindowEditor.

Simplify & Generalize stepping code, introduce stubs for reverse
debugging. When user switches to another thread than the
event-triggering one in a stopped inferior, explicitly tell
the engine to continue that one.

Improve fixDumperResult to fix missing types (out of scope items)
and call it from expandPointerDumpable.

Remove unused exec functions, replace by calls to the respective
stepExec/nextExec, which sets the code level automagically
according to the OperateByInstruction action.

Adapt to new state model, make sure session is always cleaned
when invoked from processExited event handler, remove calls
to DebuggerManager::notifyXX().

For device debugging. Move both from namespace
Debugger::Internal to Debugger, export them, give
DebuggerManager a static instance().

Intergrate per-engine state tracking with the global running/stopped state.