This addresses QTCREATORBUG-15426 for the case of well-known simple
types (int, float, ...). Retrieving 1 mio ints takes now ~15s instead
of ~140s.
Task-number: QTCREATORBUG-15426
Change-Id: Ifb7c957d60504e23147d499d8bb9bbd41c5a53ba
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Function-local struct definitions produces types 'names' like
'QList<main(int, char**)::SomeStruct>'. Remove the part up to
the '::' and rely on local lookup.
Task-number: QTCREATORBUG-15400
Change-Id: Ic73005f294c5f62c504b2d7ded4dd2716d88f2e4
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Property values are stored now using a QVector, not a QList.
Change-Id: Ifc8d6b856d2aff5ce4565c43111132f2a73460df
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
With C++11 we use std::atomic in the ref counter, giving yet
another possibility of structure member names. Instead of
trying to keep up with that, access the binary data directly.
Change-Id: Iabea831689f58f83150287218bb015c824e888ca
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
The objects have been reported as 'having children', but
when asked for them nothing was produced, leading to an
endless loop (until the next 'step' command).
Change-Id: I3bbfe10a7ed980b27a5fca1edfe44dd8983b8f7b
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
- Remove old experimental native mixed approach.
- Move some common stack parsing to Stackhandler.
- Mark gdbbridge.py debug output explicitly to remove it
from actual reponse handling
New native mixed needs QtDeclarative changes and
QTC_DEBUGGER_NATIVE_MIXED=1 for now.
Change-Id: I09eed1da51cea878636d36756015b7bfaed34203
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
The module doesn't exist in Python 2.6, so we can't use it
unconditionally if 2.6 is to be supported again.
Task-number: QTCREATORBUG-15076
Change-Id: Ic91cdf451ce846d892cd914bf850a4fd4b18c488
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
... instead of the custom mechanisms to funnel breakpoint ids, cookies
and similar through individual bridge functions.
Change-Id: I2b6e906078eefdbcd4ffa166dc1881a1d0f7fb51
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
While the formatted value is usually all that's wanted,
having access to the individual chars is handy at times.
Change-Id: I53b0d0ccfe4289b9b54a1caced4e0bd5ac66d9fc
Reviewed-by: Niels Weber <niels.weber@theqtcompany.com>
- Move DisplayFormat from watchhandler.h to debuggerprotocol.h
- Add/update a few comments about the use of the protocol enums
- Make decodeData take a DebuggerEncoding instead of an int
Change-Id: I50bed70a5da2e94da46e894bf9136bc14c9a1b23
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
... by re-using the logic we had for std::string already.
A plain char[] value did not react to Change Display Format,
Latin1 was used unconditionally.
Also rename putStdStringHelper to putCharArrayHelper.
Change-Id: I01fdf796ff49a4c99ead7b9b46274684e18e962b
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Regression since 9130cdfc05, leading to "<unavailable>"
instead of "<at least %1 items>"
Change-Id: I89166e50f27c568009ff7aeabb5adf622e303789
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
The null delimiter was not reliably found due to iterating over the
wrong positions in the string.
Task-number: QTCREATORBUG-14826
Change-Id: I3a3f2fca84648b54b12fe5fae921ce6311d4f1b2
Reviewed-by: Niels Weber <niels.weber@theqtcompany.com>
GCC 4.9.1 + GDB 7.8 (and likely more combinations) produce
sizeof(QProcess[3]) == 0, making the dumpers report no child
entries for the array.
To work around, try to parse the array size from the type name itself.
Task-number: QTCREATORBUG-14755
Change-Id: I1f17d5d6fde6bc5fe66d2ad4a73f77185503f9ab
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
... when passing data from the dumpers to the GUI. This reduces
the need to guess whether a value contains a translatable string.
Change-Id: I5e2210b8d028bd71f0087a2ba5c7c5b04331b882
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
It's an implementation detail of the gdb side (and would not
be there if gdb.lookup_type were faster)
Change-Id: I135c32a1c9132ab771c8fab5abfd04fbb308891e
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
This is not a complete solution (that would need to pass information
on typedefs and artificial types like "QVariant (QVariantMap)"
between dumper and gui, but allows switching to compact display
for individual items.
Change-Id: I47d15356e5decf9411af4acc8fa7a0f0cfbd045f
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
This partially reverts commit e33da6a6b3.
GDB seems to hang while executing str(some_pointer) in some setups.
Task-number: QTCREATORBUG-14374
Change-Id: Ia678533b9122e505b21ef6659d404239995ddb81
Reviewed-by: Orgad Shaneh <orgads@gmail.com>
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Older versions of GDB (~GDB 7.4 on 32 bit) will cause segmentation
faults in inferior calls due to misaligned %ebx values in an SSE
call in qstring.cpp:findChar.
Change-Id: I44492106080f12e645f9d57828438ec70fd66ca6
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
In the fallback case now really go back to explicit object-at-address
construction. LLDB on Linux doesn't seem to like implicit pointer
arithmetics.
Change-Id: I3be6b0b21b35b9b9141440d4a11e9e889bb2f4bb
Reviewed-by: hjk <hjk@theqtcompany.com>
The field checked for does not even exists in lldb.SBValue.
Change-Id: I062ccf62da2e2ea3f011d0b841cba0e87f7c6fa9
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
'bytes' has special meaning in Python.
Change-Id: I9ff7c56488c9d29e42b2c7a12e40b45815be2a76
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
This makes the GccExtension dumper test pass with GDB
Change-Id: Ia1319e878416bea8b162a1a9eef1271cb7bebc63
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
This practically removes any functionality beyond plain plot display,
but does that at least reliably, cross-platform, without dependency
on 3rd party python packages.
Change-Id: Iaff2f78595394522f32264c642df20dd48b83f8b
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Move common code to dumper.py and debuggerengine.cpp and
fix it there.
Change-Id: I20d91d1aa7400fbdb27938c10cf40c8f6019df0a
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
The current setup (dumper->gui: list of descriptions,
gui->dumper: index in list) is fragile and not easily
i18n'able. Go with an enum based approach now.
Change-Id: Ie78c596065a8b2ba87ad725274da29d4be3a6da4
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
... and further unify GDB and LLDB code paths.
Change-Id: Id89f3804c53190c4888082891fd3c3c55eceac84
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Was broken after making the handler functions members.
Change-Id: If14cbad777323d48000641fb38a01eb59503d297
Reviewed-by: Orgad Shaneh <orgads@gmail.com>
Move reusable Breakpoint enum to base, use 'theDumper' uniformly
across backends as name for the Dumper singleton. Signal
engine setup only if the Dumper successfully initializes.
Change-Id: If1b6be9d054f249d0d46061e4949cbecf4fd09cb
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
Instead of calling a GDB extension command invoking a function
on the well-known Dumper instance, call the Dumper function
directly. This also makes the code more similar to the LLDB side.
Change-Id: I4b23177eb72a904721b63c578ce7fbfe204f02a2
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
In preparation for re-use through LLDB. Not yet functional there.
Change-Id: Ia924f1499a1bbaebf3dccbe15152f324058b1677
Reviewed-by: hjk <hjk@theqtcompany.com>
Breakpoints are hit, stack frames are identified as JS or native.
No further data yet.
Change-Id: I84a02422fd36dc7645003114dd8519bedd913c06
Reviewed-by: hjk <hjk@theqtcompany.com>
