- added a function which tries to find the executable belonging
to a core file if the extracted path does not exist
- tries to find binaries with a relative path
- tires to remove commandline arguments, since the core
file does not store which part is the exe name (which can
contain spaces), and which are arguments
loading /QT/core which was created by "/QT/bin/exec w.." by calling
"./bin/exec with spaces" arg1 arg2 now finds the binary
==== will be split into a separate commit, included to show the
second use of the new function
LoadCoreDialog: change when ExecFilename is set from core
- only set the name if the exe exists
- if it does not exist and the previous entry does not either
then insert what was extracted from the corefile
Change-Id: I4a9846761c91ed976f3ba38a7dc756fc30ed179c
Reviewed-by: hjk <hjk121@nokiamail.com>
- the string in the gdb output by which the exec file is identified is
subject to localization. Overriding the LC_ALL variable fixes this.
Change-Id: If839ed1daec8cbc0b6a5fefcf28a4eef0423a645
Reviewed-by: hjk <hjk121@nokiamail.com>
It's unconditional nowadays, so we don't gain anything but code path
complexity from defering it.
Change-Id: Icc7d65835d1b2b76ecb64601822e0d70fd8f9f6e
Reviewed-by: hjk <hjk121@nokiamail.com>
A lot of our build system files specify unneeded include
paths. These roughly fall into the following categories:
a) Paths that are already set in more general files
such as qtcreator.pri.
b) Paths that serve no purpose at all, possibly
left over from earlier versions of the project.
c) Paths that act as workarounds for wrong include
statements of the form '#include "xyz.h"', where
xyz.h is not in the same directory as the including
file.
This patch removes such path specifications and fixes the offending
include statements from case c).
Tested on Linux, Windows and OSX with qmake and qbs.
Change-Id: I039a8449f8a65df0d616b4c08081145c18ae4b15
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@digia.com>
* Delete process when finished
* Handle unpack failure
* Support lzop on Windows (if available)
Change-Id: I8d95b77922c8b22b5874dffd8f935ff184902e12
Reviewed-by: hjk <hjk121@nokiamail.com>
This enables access to object-specific plain gdb pretty printers
which are not available before startup.
Change-Id: Icc8cbec177825d4d1adb1957a618abb33cbbf319
Reviewed-by: hjk <qthjk@ovi.com>
Fix gdmi escape routine by using a traits class to adapt
QByteArray/QString.
Change-Id: Ic6d2a0c927e4613aec9f69095bbb04a393e9edb9
Reviewed-by: hjk <qthjk@ovi.com>
- Compile with QT_NO_CAST_FROM_ASCII.
- Remove single character string constants.
Change-Id: Icece98619b6c30e047d3fce00e6ae74bbcd53c67
Reviewed-by: hjk <qthjk@ovi.com>
The name of the binary is now read using the elfreader instead
of attaching gdb twice.
Change-Id: I7a75bc64278c4f43e7085d2e479b53173602c966
Reviewed-by: hjk <qthjk@ovi.com>
This allows adapters to take all responses from the startup commands
into account.
Change-Id: I295cb211a4b69cfb8c59b34030aaee8120ffe98e
Reviewed-by: hjk <qthjk@ovi.com>
Getting the #include directives ready for Qt5. This includes the
new-project wizards.
Change-Id: Ia9261f1e8faec06b9285b694d2b7e9a095978d2b
Reviewed-by: Eike Ziller <eike.ziller@nokia.com>
(except gdbmi.cpp, name_demangler.cpp). Remove some unneeded
conversions, change some maps to take QByteArray keys.
Change-Id: I010f1251998a441fe5c8c87901b1e0c277c0391c
Reviewed-by: hjk <qthjk@ovi.com>
Warn if the condition fails, but otherwise don't change the execution
flow.
Change-Id: Id7b14c745109b66960add967b2a4ef8d31e1a546
Reviewed-on: http://codereview.qt.nokia.com/2389
Reviewed-by: Eike Ziller <eike.ziller@nokia.com>
Use proper QByteArray member in GdbResponse instead. Less magic, and
does not break array-style reponses where all children are assumed to
be of the same "type".
Change-Id: Ief1efdeb0412f59836e1b2d6f4cb751ddef3216e
Reviewed-on: http://codereview.qt.nokia.com/1184
Reviewed-by: Qt Sanity Bot <qt_sanity_bot@ovi.com>
Reviewed-by: Eike Ziller <eike.ziller@nokia.com>
The breakpoints are now (fairly) tightly guarded by the BreakpointHandler.
Engines and Views are only supposed to refer to them by id. They also have
individual states now. The breakpoint data is split into a "user requested"
"fixed" part in BreakpointData and the engines' acknowledged data in a new
struct BreakpointResponse.
TODO: Move m_state and m_engine members to BreakpointResponse. Fix regressions
in the marker handling.
