Lukas Diekmann

If I understand you correctly, then you are suggesting to reverse both the order of the assembler instructions as well as the bytecode they emit. For example, if I have...

I wasn't suggesting it was your patch. :) I just rememberd it working before I pulled yesterday evening, but I'm not sure what commit I was at before.

This is the segfault I get with gdb: ``` Program received signal SIGSEGV, Segmentation fault. 0x0000555555563f3c in rgtk::gtk::signals::draw::trampoline (widget=0x555555870210, ctx=..., signal=0x0) at /home/lukas/research/rgtk_fork/src/gtk/signals.rs:90 90 let t : &'a Box =...

And here's a backtrace. Anyone an idea what I should be looking for? ``` #0 0x000055555556512c in rgtk::gtk::signals::draw::trampoline (widget=0x555555878210, ctx=..., signal=0x0) at /home/lukas/research/rgtk/src/gtk/signals.rs:90 #1 0x0000555555564ffe in gtk::signals::draw::trampoline::hac039c50530d7131oxe () #2 0x00007ffff77161ee...

Since the stuff I am currently working on depends on this, I wouldn't mind fixing this myself. Unfortunately I don't know where to start.

Well, the segfault seems to suggest it's this line (signal.rs:90): ``` let t : &'a Box = ::std::mem::transmute(signal); ```

Thanks @gkoz and @GuillaumeGomez. :+1: Commenting the `Drop` implementation worked. I can live with a bit of memory leak for now if that means I can continue working. :) What...

Ah that makes sense then. Thanks for the explanation. :) So that means removing that `Drop` implementation is the proper fix then (in case cairo::Context is never used anywhere else)?...

Yes, I believe this is due to the shrinking/expanding feature of automatic language boxes (which we didn't have before so this wasn't an issue before). The problem are whitespaces within...

Hm, that might work. I will give this a go and see how it holds up in practice.