@Pl45m4

Thank you for your feedback, I'm fairly experienced with plain C++ but not very with Qt or using Valgrind with external dependencies so I was a bit confused about what was happening.

I'll keep that in mind in the future!

I have encountered a similar problem, which has not been resolved. However, your example does not illustrate this issue.I test the example ，when myModel = newModel，print myModel.length =0

There is no leak. How do you measure? Please provide a minimal, compileable example of the problem.

@vicky_mac said in Memory leak in QML example project:

Is there anyway we can prevent that?

No, as already replied in your other thread asking exactly the same, which you have not referenced from this topic. But you can keep asking the same. Or, you can look through the source code, find the exact point of the leak you claim, and fix it.

Apologies, that was https://forum.qt.io/topic/152049/stackview-is-cause-of-memory-leak, you do not seem to be the same author, though they look similar.

I also notice you are using Qt 5.12. Qt Company is not going to be addressing anything in that version.

Hi,

See if this provides some explanation (read the comments as well):

StackOverflow: Qt6 QLabel / QPixmap excessive memory usage

@Axel-Spoerl said in A Strange Qt and postgresql problem ! when using Kaspersky:

If everything is configured properly and you use pgsql > 10.7, the only idea I have is that Kaspersky (partly) blocks network traffic coming from an executable, which it doesn't know. But that's where my knowledge ends and you have to turn to the Kaspersky support.

Thank You so much bro <3 for great support
by the way i am using postgresql version 14 the latest one

i agree with you there's a problem on kaspersky they partly affect any unsigned software when connecting to postgresql ... (kaspersky set my tested softwares as (low restricted which is the best group when application doesn't have digital signature) in application control

Low Restricted. This group includes applications for which the following conditions are met: Applications are not digitally signed by trusted vendors. Applications are not recorded in the trusted applications database of Kaspersky Security Network.

@BigBen said in Potential memory leak in QSignalMapper : Qt version 6.3.0:

connect(myMapper, SIGNAL(mapped(int)), this, SLOT(up(int)));

I'm pretty sure that this:
connect(myMapper, SIGNAL(mapped(int)), this, SLOT(up(int)));

should not be done in the loop.
There should only be one connect of the mapper to the receiving object

@leonardoMB

https://doc.qt.io/qt-5/qwidget.html#setParent https://doc.qt.io/qt-5/qlayout.html#addItem (or addWidget)

Hi @kiozen, I am using Qt creator on windows. Please let me know how to run a application with ASAN on qt

@Christian-Ehrlicher Thanks for tracking that down. Hope Qt gets it fixed soon.

@Roysten-Rigobert Why don't you create menu/actions at once by making menu as member variable?

and in contextMenu(QPoint) slot call only the menu->exec(mapToGlobal(pos));

@SpaceToon said in Replacing QList of objects with QLIst of pointers to these objects:

why in my solution with the pointers an exception occurs.

Because the pointer you store a dangling - the object behind is already deleted when deviceDiscovered() is finished (or a little bit later, don't know the code which calls this function).

You need to add extra logic in your prog.

Your problem seems very similar to the Copy/Cut/Paste logic. You don't know if the paste command is valid because you don't know if the user press copy command before that.
Well, is up to you to register the states of this commands.

So you need to implement the logic suitable to your case, for example, if the list of items is empty, obviously you cannot edit it, etc.

About the interfaces (as others said), create them on the fly could be tricky and error prone. You can do it if only minor changes are to be made (hide/disable a few button/options)
If the interfaces are very different, better way to create a unique persistent widget for each, and put them in a StackedWidget.

Good luck :;)

@legitnameyo
Although what @Kent-Dorfman says about stack-allocated destructor auto-closing is quite correct, I would always put in an explicit close() as soon as I am finished with any file, personally. It's good practice, especially for potential future code changes. You never need flush() on read, and you don't need flush() on write if it's immediately followed by close().

@fcarney I can't beleive I missed your question. Comparing with Heob, Deleaker provides more features like taking snapshot at any moment, and comparing them later. Besides heap allocations, Deleaker also tracks COM/ActiveX/GDI/USER32 leaks etc.

@aha_1980

That's what I thought. If I manually start valgrid it runs fine, it produces a report as well.
I just wanted to make it work with Qt because it looks more convenient.

I'll try updating Qt creator later on.

Thanks