Tony Wasserka

This is a handy feature that I found myself looking for previously too. I don't know what's blocking Catch2 from supporting it currently, but a workaround is to use dynamic...

> array access on x86 for non-constant indices is just... stupid?? What exactly is it that's stupid?

> I don't know why all of this is necessary to access a 1 dimensional array. Here's your new best friend: [Inline-constexpr variables](https://godbolt.org/z/G4Wa1cfc5) 😉 ``` unsigned short CalcMask(unsigned long): movzx...

The design conflicts observed in this PR are one of the reasons why for ongoing projects I've moved away from the `union`-based approach. My current design represents bit fields as...

As mentioned in other channels, what I need to review your ThunkLibs PRs is better separation of changes and documentation. The bare minimum needed is a rudimentary explanation of what...

> Export only symbols defined on the interface files from the guest thunk, and hide all others to avoid exposing thunk lib internals to the guest. Sounds good. Did exporting...

> Yes, of course, the PR was just to propose the ideas and make sure we are aligned. Will chop them as non-rfc ones. Great, cheers! > As the interface...

> No, because this happens in the relocation phase, not during dlsym. Think of it as, host GL library doesn't return &glGetString, it returns an internal, unnamed symbol. > >...

> guest dynamic linker Ah, I was mixing up a couple of things there, but yes of course we can't actually override the `dlopen` flags for the guest linker easily...

As a general remark, I'll quote my comment from #1734: > As mentioned in other channels, what I need to review your ThunkLibs PRs is better separation of changes and...