sf::Text::getLocalBounds causes segmentation fault?(SOLVED)

[lumaio]

Calling getLocalBounds on an sf::Text object causes a crash.

minimal example:

#include <iostream>
#include <SFML/Graphics.hpp>

int main()
{
  sf::Font f;
  f.loadFromFile("font.ttf");
  sf::Text t;
  t.setFont(f);
  t.setString("some string");
  t.setCharacterSize(16);
  sf::RenderWindow window(sf::VideoMode(1280, 720), "example", sf::Style::Close);
 
  while (window.isOpen())
  {
    sf::Event event;
    while (window.pollEvent(event))
    {
      if (event.type == sf::Event::Closed)
        window.close();
    }

    sf::FloatRect width = t.getLocalBounds();
    window.clear();

    window.display();

  }
}
 

My OS is Windows 10 Pro 64-bit. As far as I can tell nobody else is having this issue so maybe I'm just not using it correctly?

[Ausche]

What is the exact error message? This looks right.

[lumaio]

That's the thing, There wasn't an error message. Just the standard windows "whatever.exe has stopped working"

[Hapax]

minimal example:

The example should also be "complete".
You left out the event code.

[lumaio]

minimal example:

The example should also be "complete".
You left out the event code.

I don't see why it matters but I added it.
Is this an issue with SFML or my own code? It's driving me nuts.

[Jesper Juhl]

I don't see why it matters but I added it.

It matters since if the example is not complete then people can't copy-paste it and compile/run it on their own machine.

[Nexus]

That's the thing, There wasn't an error message. Just the standard windows "whatever.exe has stopped working"

Run it in the debugger and in Debug mode. Then you get a more meaningful error message and see where exactly it crashes. Once you've done that, provide us with that information.

[FRex]

Is it crashing in any way with the event loop too?
I'm asking because your "crashing" is normal without the event loop, without an event loop in it your program is literally not responding to windows so that message appears.

[lumaio]

That's the thing, There wasn't an error message. Just the standard windows "whatever.exe has stopped working"

Run it in the debugger and in Debug mode. Then you get a more meaningful error message and see where exactly it crashes. Once you've done that, provide us with that information.

Best I can do is GDB because visual studio won't link properly for whatever reason.

(gdb) run
Starting program: Z:\Projects\test\test.exe
[New Thread 5868.0x2030]
[New Thread 5868.0x263c]
[New Thread 5868.0x211c]
[New Thread 5868.0x2a28]
[New Thread 5868.0x1c70]
[New Thread 5868.0x21cc]
[Thread 5868.0x1c70 exited with code 0]
[New Thread 5868.0x2aac]
[Thread 5868.0x2aac exited with code 0]
[New Thread 5868.0x2a80]
[New Thread 5868.0xa10]
[New Thread 5868.0x54c]
[New Thread 5868.0x2674]
[Thread 5868.0x2a80 exited with code 0]
[Thread 5868.0xa10 exited with code 0]
[New Thread 5868.0x1230]
[New Thread 5868.0x26a0]
[New Thread 5868.0x27d0]
font loaded..

Program received signal SIGSEGV, Segmentation fault.
0x0000000068eec64c in sf::Text::ensureGeometryUpdate (this=0x8000) at Z:\Source\SFML\src\SFML\Graphics\Text.cpp:350
350         if (!m_geometryNeedUpdate)

Is it crashing in any way with the event loop too?
I'm asking because your "crashing" is normal without the event loop, without an event loop in it your program is literally not responding to windows so that message appears.

I checked, this is not the case.

[Hapax]

You left out the event code.

I don't see why it matters

without an event loop in it your program is literally not responding to windows so that message appears.

That. That it why it matters 

I checked, this is not the case.

Are you saying that it doesn't have a fault when you do have an event loop?

[lumaio]

I checked, this is not the case.

Are you saying that it doesn't have a fault when you do have an event loop?

It still crashes with an event loop.

[Nexus]

because visual studio won't link properly for whatever reason.

What does that mean? Again, it would be helpful if you could provide an actual problem description instead of a vague statement. This may indicate a configuration problem, and it's useless to debug SFML when the error is simply linking the wrong binary...

In case of doubt, try to rebuild SFML and link your own version. Make sure that you follow the tutorial steps precisely and don't mix any configurations or compiler/linker flags.

[lumaio]

Building with visual studio and release libs results in the same crash but using the debug libs works fine.
The same can't be said for mingw though, both release and debug have the same issue.
I'm not sure what info from the visual studio debugger you want me to give you, gdb gave me more info than I got from vs.

[lumaio]

So reading this thread back I wasn't clear at all what my config was which is probably confusing so here;
I'm using mingw 5.1.0 and compiling with clang 3.7.0.
I was able to get the code in the OP to work if I compiled in visual studio with the debug libs.
The crash happens at line 350 in Text.cpp

Code: [Select]

if (!m_geometryNeedUpdate)I noticed that sf::Text::ensureGeometryUpdate is called when rendering as well but it doesn't crash when I do that, only when I try to get the bounds (both getLocalBounds and getGlobalBounds).

[Estivo]

Building with visual studio and release libs results in the same crash but using the debug libs works fine.
The same can't be said for mingw though, both release and debug have the same issue.
I'm not sure what info from the visual studio debugger you want me to give you, gdb gave me more info than I got from vs.

I checked it on Win 10 64 bits with SFML 2.3 32bit and both (debug and realease) and VS 2015 (v140) are working fine.