newby alert:
I have started my 10 day trial and I am having issues figuring out how to get it to find what I'm looking for. I have a fairly complex single threaded Java app that I'm trying to get to never GC. I pool all my objects. However I must have missed a few things. I'd like to find out where I'm releasing the last reference of the objects that are going back to the heap. If I cant find that, it would suffice to at least know where the memory is being created that is later being GC'd.
This is what I'm doing so far:
1. click on Live memory / Recorded Objects
2. Click record and select "Garbage collected objects, All allocations"
3. Click on "Run GC"
so now a bunch of stuff appears. I was hoping I could simply double click and it would give me line numbers and fun stuff like that. Similar to what I got when I would run HPROF.
I right click on one of the items and select "show allocation tree for selection"
Still no line numbers... it does give me the method name though. If the method is large it's kinda complicated to figure out where the memory is being created.
Also when I stop recording, is there a way to clear everything in the recorded objects view so I can start over with a new recording?
In the profiling settings you can switch on line number resolution.
Also when I stop recording, is there a way to clear everything in the recorded objects view so I can start over with a new recording?
When you start recording again, all previous data is cleared.
You can try this:
Click "Live memory - Recorded Objects"
Choose the Objects you want to watch then right click "Show Selection In Heaper Walker" and choose "Show in current heap dump"
Choose the Objects you want to watch in the new Heap Walker window and right click and choose "Use Selected Objects" and choose "Allocations". Then you can see the allocation tree of the Objects, by read the method code you can locate the line number you want.
Related
I hope I can explain my specific annoyance clearly enough:
Quite some time ago, IntelliJ introduced a feature in the run console to parse the content of the output and would try to print a nice tree structure on the left. However, its default behavior of jumping to the error was very annoying. To illustrate my point:
When it starts normally, the full console output is available in the right panel. However, when things happen, or, in my case of gradle run, when I simply stop it, IntelliJ would jump to the specific block of output only with the error:
Note, in the second screenshot, that there is no scroll bar on the right side of the right panel, and thus I cannot scroll up to view the actual run logs. I have to click the left panel's top element in the tree to be able to view all logs.
The locking to a specific block of output is nice to have, but I am happy without it too, as before it was introduced. But the default jumping to that block is annoying and irritating. Almost in all cases I started and ended a run like this, I have to click the left panel to view the whole log -- this "Build cancelled" error block is of no use at all. And when it's jumping to an actual exception, majority of the time I'd need to look around its context for the cause, which requires me to re-orient myself in the log flow where the block was at. This is very counter productive for me.
I have been assuming IntelliJ should provide an option to either not jump to the error by default (leaving the full log available in the right panel), or to disable the whole console output parsing as a whole.
I have not been able to locate a setting for either of this. Does anyone know if that's possible?
We are working to fix that. Please vote and follow https://youtrack.jetbrains.com/issue/IDEA-241844
I have written an app that makes extensive use of custom right-click menus on an Access form. The code works great and the user loves it, but lately I am having trouble making it work properly.
In earlier versions of Access it worked well, but newer version seem much more limited in how many items can be put in such menus. The documentation is silent on the matter, and nobody in any newsgroup has had any useful ideas, but I regularly get random error about stack space, out of memory, and general lockups when populating the menus. Doing a C&R used to help, but now even that is not enough, and some menus I can no longer populate at all.
I tried building an app that just built menus until it crashed, to get some idea of what the limits may be, and I am well below what that indicated, but the experimental app had nothing else, while the real app has a great deal else.
Is there any information on how much stuff can be put into these menus, and what the menus share space with? There may be something I can do another way to make more room for the menus. I tried moving all code out of the form, leaving only event stubs that called routines in standard code modules, but that did not help.
And how are they stored/activated? The app is MUCH slower to load when it has these menus, even though no code is running on start-up.
********** Edited to add this:
I use VBA to create a menu, like this:
Application.CommandBars.Add "RCStat", msoBarPopup, False, False
then add it to a control. like this:
Application.CommandBars.cboStat.ShortcutMenuBar = "RCStat"
I add controls (only popups and buttons) like this:
Application.CommandBars.Controls.Add(type:=msoControlPopup)
Application.CommandBars.Controls.Add(type:=msoControlButton, Parameter:="StatKod = 77")
It runs perfectly and the menu items work exactly as expected, except that it bombs after adding some number of controls. It doesn't seem to matter where I add them, just the total number of added controls hits some undefined threshold, and the app crashes.
I got the original code from Getz, Litwin and Gilbert, 2000 edition. Back then, it worked great. But as the versions advance and the app accumulates data, it is becoming less and less functional. However, there are only around 10,000 records, and the app itself is less than 100MB - nowhere close to any of Access's upper size limits.
Pete,
I've done quite a bit of work with shortcut menus, and created the Access Shortcut Tool about 5 years ago, but have never attempted a menu with so many controls although some have 3 or 4 levels.
I am not aware of any restrictions on the number of elements in the commandbars collection, but I find that shortcut menus with too many options, like lists and combo boxes with too many items, are difficult for users to navigate. I generally break these up into segments and use buttons in the form header to display the appropriate menu. Sorry I'm unable to provide anything more helpful.
Dale
We have a commercial product, Total Access Components, that includes as one of its 30 components a right click popup menu that can include icons and font styles.
Here's the info for the popup menu control: https://fmsinc.com/MicrosoftAccess/controls/components/popup-menu/
There's a free trial if you want to try it.
I wrote a Pharo program that generates my daily task non-stop.
The program itself works fine, however I always need to instantiate my object in the playground to run it. Is there some other way of doing it automatically without having to create an object and send a message to it?
Yes, you can. Save the image after you have instantiated your application and closed everything else. Then just start pharo from the image and you will have your application started.
If you want to do it more production wise you could use pharo-launcher.
For more detailed information you could squeak wiki which you can adjust to Pharo.
I would do one of three thing here:
Add an item to the World menu, so you can bring up the menu and select your task. See a Stackoverflow Answer on that same topic.
You could write some triggering code in a .st file on your file system, then use StartupPreferencesLoader to load it on startup.
Create a window morph with a button that, once pressed, runs your code. Open the window, quit and save image changes. Never close the window.
I'm new on this, sorry for bad usage of terms or overextending an explanation. I'm learning code languages and way i found to bring it to my world so i can learn it better, was coding for/with games i play.
When a window close on the game a function needs to be called, when the character move or you pick an item, everything has a command, function, process or some value of an address change and etc... What i wanted to know is if there something that shows me on real time every call, every value change, address value change, etc...
Nowdays i have to reach some value address by CheatEngine, changing the value till i find the correct address. With this kind of thing i would have a list off things that is happening right now, and a "log" of the past things, then i go to the exact time that i did something, so i would have to look on that peace of the list and discover what did my "something"
Click on a button;
Check on the real-time thing what happened at the time of the "Click
on a button" process;
Discover what call was responsible for that and what it did;
Now i can code something that do what "Click on a button" do, without needing to actually click on that button;
I have seen it somewhere, thats why i'm asking here, if i'm totally wrong and this doesn't exist, i'm sorry, i will delete this post.
I believe a Debugger is what your looking for.
The Debugger in Visual Studio does exactly what your describing, you can see the real time values of your variables and objects at any stage in your program.
You can set 'Break Points' that will halt your code wherever you put them, allowing you to inspect your variables.
You can also 'Step' through your code, which will go line by line through your code, stopping at each line to allow you to inspect.
Here is a good page for an overview of debugging in Visual Studio 2017:
https://msdn.microsoft.com/en-us/library/y740d9d3.aspx
I'm trying to get better at using the debugger rather than just always printing my debugging message out via NSLog. This morning I thought I had a great use for it - to find how the changed values of a KVO dictionary. But when I set a breakpoint just after the appropriate code and ran with debugging on, all I saw in the debugger was this in the summary:
{(int)[$VAR count]} key/value pairs
and expanding the item never gets to any key or value.
But when I print out the "change" dictionary via NSLog I get the key and the value:
[timestamp] appName[3643:707] observeValueForKeyPath new filePath change: {
kind = 1;
new = foo;
I guess I have two questions:
Is there a way to get dictionary keys and values with the Xcode debugger?
Is there a set of rules of thumb for using the debugger vs printing log messages?
I'm (still) using Xcode 3.2.6.
Thanks!
EDIT: I realize that "foo" is not a valid file path. At this point I'm still testing sending a value around via KVO.
Set a breakpoint on the spot you want to check the value of the variable and run your application. When the breakpoint gets tripped put your mouse over the variable in the debugger. A little popup should come up, move your mouse over the arrows on the left side and another menu should popup. Click on "Print Description" and it should display something similar to a log in the console.
It is just a case by case method but a lot of it is just personal preference. You will just have to find what works for you.
If I am going to check a variable a lot at a one specific point, I will go with a log. It is usually the quickest way to see if something is working right
If I am checking how a part of the code affects a variable I will use the debugger and step through the code. It is also useful for checking the values of lots of variables in a section of code.
You can using a future set, which defenetely working in Xcode4
When u setup breakpoint, where is a option to don't stop execution of program and write to log anything what u want.
I'm not sure if it working well in Xcode 3 bcs I familiar with 4 but here is work.