I just started a new task at a "lower level" in the platform stack, and I'm getting started with windbg. I'm so far quite happy with the pure power of the debugger. However, I wish it would just save my session default, like the VS debugger does. What I want is that whenever I ".restart", or re-open windbg, it works just like I left it: same bp's, same sxe state, same files open in the same places, etc.
I know about "save workspace" which seems to do what I want, but it's manual, and I have to do it every time I make a change to the workspace state.
Is there a way to just have windbg do this automatically?
It should prompt the first time you close the session and ask you if you want to save your workspace, there is a checkbox like the image here.
If you click yes this time and the box 'Don't ask again in this WinDbg session' then it will automatically save your workspace, similarly you can also clear the workspaces if it's erroneously saved some breakpoints or paths that you are no longer interested.
Also you can set this in the options like so:
Microsoft NTDebugging Blog. Uncovering How Workspaces Work in WinDbg.
Related
I'm doing an automation script for installation wizards using AutoIt. I'm trying to handle window changes in some way.
Can some one explain how these GUI's work?
When I click on the Next button it looks just like the components in the GUI is beeing changed. Is this tha case? Or is a new window created and the old destroyed?
I've noticed that the process ID is the same for all windows.
I'm sure there is some way to know which "state" the GUI is in, or which step?
By the way. All the windows has the same title.
Thanks
/Anders
This will be dependant on the program you are automating.
The easiest approach would be to look at what changes in the GUI between stages, likely candidates are if there is a label that is giving instructions for that step, or a button that has text changing (e.g. if the button says "Finish" then you know your at the end).
Most installer programs have child windows for grouping the controls of each stage. These are typically implemented as dialog resources (as can be seen when using something like reshacker on them). So although the window remains the same, the panels are being created/destroyed as appropriate. This is a very neat method of doing it, for the obvious reason that you don't need to have to code to create/destroy a lot of controls. Resource created dialogs don't have nice class names like windows sometimes do though, so this may not be a reliable way to check the state.
I very rarely close down Visual Studio intentionally. The project I'm working on is so ginormous that it takes about 15 minutes just to load it up, so it's been my practice just to leave the solution open overnight so that it'll be ready for me the next morning. The only time I have to restart the program now is if it crashes - which - happens. Every once in a while I'll do something it doesn't like and it quits on me. I accept this. I'd love for it to be bullet-proof, but I hardly expect that.
The downside of this is that apparently, any settings you change in the program while it's running (fonts and colors, keyboard shortcuts, auto-formatting settings, basically anything in the Tools/Options dialogue) are only saved when you quit the program legitimately. These settings are not saved when it crashes. So I find myself having to redo any settings changes I made every single time.
First off - why the heck doesn't it save the settings when you change the settings? How hard would that be? It's a certain level of hubris to take it on faith that your program will always close normally and not bother saving state until the shutdown process. (And not having it save settings in an abnormal shutdown process either) I'd get fired (or at least reprimanded) for writing code that did that, so how does Microsoft get away with it? And is there some way of forcing it to save settings without quitting and reloading?
As far as I know, the only thing you can do in your situation is to export your settings directly after you change them, and import them after a crash. This way your settings are safe, and you don't need to restart.
Example: i'd like to always see "CPU Registers" and never see "Scripting Console".
Using Code::Blocks, how can i do this, regardless of the project i use?
You realize the CPU register window won't show anything unless you're actually debugging right?
Anyway, to save a perspective in C::B, setup the IDE to the layout you want and just save it. In your case, you would do something like the follow:
have the debug toolbar visible. Enable it by going
to View->Toolbars->Debugger.
display CPU window from debugger toolbar. Find 'Debugging
windows'->'CPU Registers'.
Hide the scripting console. Go to View->Script Console and uncheck.
Set the layout to your preference and save. Go to View->Perspectives->Save current. Give it a name.
The next time you start C::B it'll use the last perspective you were on when you quit. So make sure you exit C::B with the new saved perspective.
What is your biggest pet peeve related to the windbg debugger from microsoft?
(note: I actually really like windbg if I ignore the unpolished UI.)
Attempting to dock a window is almost always the wrong kind of dock the first time until :I move the mouse just right. Why can't it have the docking cues that VS2008 has?
The ridiculous behavior when you attempt to use click-drag to select text on a line that is wider than its physical window.
The pieces of the history window that I need to copy/paste into bug reports are frequently wider than the physical window. I've gotten so used to the triple-click workaround that I find myself attempting to misuse triple click in other (well behaved) applications.
Key presses are ignored while the focus is in a source window. It's not like you can edit the source code from inside windbg. At least there's the Alt-1 workaround.
How insanely slow .kdfiles copies new binaries across the 1394 connection. It can take up to one minute for a large dll.
Not being able to switch from output window to command window using a key press, I read that 'Alt+1' should work but it doesn't always so I always have to resort to using the mouse. Also sometimes it doesn't remember when I tell it not to ask me everytime if I want to save the workspace when I quit.
I am currently working on a program to immediately clear the list of previously-run-commands which appears in the Windows Start -> Run dialog. The procedure for clearing this list by removing the HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\RunMRU key is well documented; however, before these changes take effect, it seems to be necessary to do one of the following:
Restart the computer
Select Start -> Shut down, and then select Cancel.
Neither of these is ideal for the task I am trying to accomplish: #1 is extremely disruptive to the user, and #2 appears to require additional user interaction.
Does anyone know how to immediately (and programmatically) force a reload of this information without requiring any user interaction, while also minimizing disruption of the user's other activities? I would like for the user's Run history to be cleared out immediately after executing my program, without requiring any further action on their part (such as using the "Shut Down" -> "Cancel" trick in #2 above) or forcing a reboot.
Or, to approach the problem from a different angle: When clicking Start -> Shut Down -> Cancel, Windows Explorer reloads the RunMUI key. Is there a way to force a similar reload without having the user select Shut Down and then Cancel?
Things I have already tried:
Monitoring the explorer.exe status using procmon while selecting Shutdown and then Cancel. I see Explorer writing to the RunMRU key, but have not been able to determine what triggers this.
Numerous Google searches along the lines of "reload runmru without reboot". Most results still recommend method #1 above, although a few suggest #2.
Limited MSDN API examination. The RegFlushKey call appears promising, but I haven't ever used it before, so I don't know if it will apply to registry information cached by different processes.
Any suggestions or other information would be greatly appreciated.
Have you tried ccleaner?
http://www.ccleaner.com/
Not a full answer to your question, but I did find a third way to trigger the clearing of the run command from this article in PC Mag.
Killing explorer.exe and then restarting it will also clear the run list after the registry modification.
I have a nasty hack for you. Show the window programatically, hide it immediately (programatically) and click cancel on it (well, you guessed, programmatically).
You might try looking for the icon cache flush API, or other ones, I wouldn't be too suprised if they had side effects like the one you are looking for.
I've seen instances where it actually works, even the F5 key doesn't work? Try this, ctrl>alt>delete then go to task manager, processes tab...end explorer.exe. Then click on file new task and type explorer.exe, then check...does that work?
Windows XP
Right click on the taskbar
Properties menu option
Start Menu tab
Customize button
Programs pane
Clear List
Click on OK
This calls a Windows API function that refreshes the explorere.exe taskbar process and also clears the list (no need for registry edits).
As far as I know, it relies on the explorer.exe process that hosts the start menu/taskbar/desktop being closed and reopened. There is no "clean" way to do this that I am aware of.
If you really need to do this without user interaction, you need to close all explorer.exe processes and relaunch one.
Here's a rudimentary C# program to do that;
using System.Diagnostics;
Process[] procs = Process.GetProcessesByName("explorer");
foreach (Process proc in procs)
{
proc.Kill();
}
Process.Start("explorer.exe");
Note that this will close all "Windows Explorer" windows open, and may or may not open an additional "Windows Explorer" afterwards.
I just tested that on Windows XP 32bit, and it did indeed clear the Run command cache.
HKEY_CURRENT_USER\ Software\ Microsoft\ Windows\ CurrentVersion\ Explorer\ RunMRU\