This is a follow-up to the work done to show errors when attaching to
processes. We were also not showing errors when trying to start a new
process using the "Debug new executable..." context menu option of the
debug view.
Change-Id: I9f68410b6d41a28df6f76f63b091a049a38b14e9
This is the first step towards cleaning up the actions.
Also make sure teh remote target provider only acts on remotes of the
type it's associated with.
Change-Id: Ifb2e39620e7d62b075c3d18dd338694071dc5720
This patch adds the "Switch Off Tracing" option to
the select reverse debugging method pull down menu.
Change-Id: Ib184c1f4c1d1db781bea4468bc8c939bd27db27d
Signed-off-by: raddepal <ravitheja.addepally@intel.com>
- unfortunately same mode is re-used for different launch group, which
have different images
- so we cannot cash icons by mode since it show wrong icon
- we also has to notify mode select that mode change event it is
"equals" so it can update launch group image
- also fixed resource leak for launch bar button
Change-Id: Iaa7b80d74963e4d0d1ccef5e6e4cd54a3ae5a4d4
There was a race condition between commandTimedOutDuringSession
finishing and the next test starting that checked for unterminated
launches (in BaseTestCase removeTeminatedLaunchesBeforeTest).
While commandTimedOutDuringSession did wait for the shutdown to be
received (ICommandControlShutdownDMEvent), that can happen a few ms
before the launch is fully terminated (fLaunch.isTerminated() == true)
Change-Id: Ib4e6655938f6b0e3ab3ce89fd06528f1b72ee04d
Signed-off-by: Jonah Graham <jonah@kichwacoders.com>
When using DSF-GDB, the handleSuspend event is received only when a
breakpoint is hit (i.e. not when it's actually suspended),
However memory changed events are received when applicable during
stepping, however the base code does not call the "archiveDeltas()" for
this event.
This change adds a call to archiveDeltas if the update mode is set to
UPDATE_ALWAYS upon reception of memory change events.
TO REPRODUCE this problem:
The issue is that only one snap shot of history is applied with the first
address / expression search, and this is used for all other comparissions,
So you can search for a variable with bigger address value as a first
memory search, and then step the code and see that variables close by
with smaller address values will be updated but not colored.
e.g.
&Test -> address 0x110
&X -> address 0x100
if &Test is used as the first memory search, and then the code changes
value of &X while stepping the code, the cell holding &X will not be
colored as expected.
Change-Id: Ibdc1fae6ed73ec2819fe9539d56f9ea9b6302535
Every time we report a problem in a declaration/expression that occupies more
than one line, the codan marks the entire line as a problem - including
everything before the node. This patch solves this inconvenient behavior.
See https://bugs.eclipse.org/bugs/show_bug.cgi?id=486610 for a complete
explanation.
Change-Id: I742cbaac8a1392676695d574355597b0cfc87385
Signed-off-by: Alisson Linhares de Carvalho <arescarv@gmail.com>
Remote connection type providers must also register as target types
if they want to show up in the launch bar. This change eliminates the
autoregistering we were doing and instead provides reusable classes
to handle the interface. Arduino does that coming up.
Change-Id: Iafb9305225f1ba3b97640bb3b15bfb671888a914
