New Dispatcher Issues

[d.healey]

Building HISE on Linux is currently segfaulting. (Using latest commits)

[Thread debugging using libthread_db enabled]Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".Programreceived signal SIGSEGV, Segmentation fault.juce::CharPointer_UTF8::isEmpty (this=0x2f7e258 <juce::juce_xmltextContentAttributeName>) at ../../../../JUCE/modules/juce_core/text/juce_CharPointer_UTF8.h:7373         bool isEmpty() const noexcept
              { return *data == 0; }

[d.healey]

@d-healey I just reenabled the Linux Font Handler (I usually have it disabled because it slows down HISE startup). And I get a bit further before the segfault

[Christoph Hart]

jupp, that's on me. Weird that it crashes though. Have you changed the HISE_USE_NEW_CODE_EDITOR preprocessor? That would explain it, otherwise...

[d.healey]

@Christoph-Hart It seems to be the fix autocomplete commit where the issue started, the commit before that works.

[d.healey]

Develop branch still segfaulting on Linux. Anything I can do to help narrow it down?

[d.healey]

@Christoph-Hart said in New Dispatcher Issues:

Have you changed the HISE_USE_NEW_CODE_EDITOR preprocessor?

Oh somehow I totally missed your question. I haven't touched that. I'm just building the unmodified develop branch straight from your repo.

[d.healey]

@Christoph-Hart Still segfaulting on Linux with the latest develop branch

[d.healey]

Maybe this is useful...

It's returning a nullptr from juce::Component::findParentComponentOfClass<hise::ComponentWithBackendConnection>()

So is there some component not being created when it should be?

Segfault occurs on line 95

[d.healey]

I've narrowed it down to this commit - https://github.com/christophhart/HISE/commit/b2471b5bbe54fe177ee43eb57a4a7839bcd0aa48

Commenting out line 118 in ScriptingPanelTypes.cpp gets rid of the segfault, not sure why though...

[d.healey]

@Christoph-Hart Any suggestions??

[d.healey]

Triggered by line 78 too. I think I'm getting close.

Shouldn't any here be a component not a mouse listener?

[Christoph Hart]

@d-healey I ran into this now after resetting the layout. It should be fixed now but let me know if it still crashes on Linux.

[d.healey]

@Christoph-Hart WooHoo! Thank you, it's working here.

[d.healey]

I'm seeing these Skipped changed() callback messages each time I hit compile in the latest version. Are they necessary and any way to turn them off?

[Dan Korneff]

@d-healey I noticed this as well. Is it actually a warning we need to be aware of?

[Christoph Hart]

I vaguely remember that there was an issue with the changed function not doing anything in the onInit callback.

However I can‘t change this behaviour for backwards compatibility reasons so I‘ve added a warning there to communicate this. So nothing has changed, but now you know what‘s up.

[d.healey]

@Christoph-Hart said in New Dispatcher Issues:

I vaguely remember that there was an issue with the changed function not doing anything in the onInit callback.

Aha yes, I brought that up a couple of weeks ago. Ok so now I will remove the useless call to changed and the warnings will go away.