Mikhail R. Gadelha

Yeah, I _think_ if we just make everything symbolic (even the incomplete ones), and rely on thrash_type_symbols to expand them should work. It should be a pretty minimal change, but...

> > just make everything symbolic (even the incomplete ones) > > @mikhailramalho I see 2 problems with making _everything_ symbolic: struct padding and initialization both examine the respective type's...

-m32 forces compilation to gen 32 bit binaries, you shouldn't use it.

That's probably why you're seeing `armv4` in the triple as clang (differently from GCC) defaults to armv4. For armv7, you need to add `-mthumb -march=armv7-a` to the command line. But...

Oh, I see. Rafael, what's the compilation error you get without this change? Em qui., 14 de jul. de 2022 às 11:44, Franz Brauße ***@***.***> escreveu: > -m32 forces compilation...

Do we really need more globals? :'(

Right, I think there are two cases that might benefit from singletons: configt and optionst (configt holds a reference to optionst?). We can at least simplify a bunch of constructors...

I also want to get rid of messaget, there is a wip patch in a separate branch but the changes are just to big to be finished quickly. On Thu,...

It's called message_interface_simplify. Em qui., 30 de jun. de 2022 às 14:35, Franz Brauße ***@***.***> escreveu: > @mikhailramalho Which branch? > > — > Reply to this email directly, view...

pthread types in linux and macos are different, that should be the reason it only crashes on linux.