Andreas Rumpf
Andreas Rumpf
Docs need to be updated. One point of ORC and the shared MM is that we can send without deepcopies.
With --mm:atomicArc it is. If it's not you `move` the references around and then you need to ensure uniqueness for this to work. IME it's easy to do that in...
I think this has been fixed with our new disabling of strict aliasing.
Too bad :-)
Works as expected. We can improve the documentation and mention ``appType == "lib"`` and similar.
> It needs (for new functionality) a stronger type system, especially more fleshed out concepts, hence the RFC it was brought up in. FWIW I had it working in a...
That's a breaking change. Instead the generic instantiation cache should distinguish between Option[uint32] and Option[Mode].
> If importc aliases were made implicitly weakly distinct it would solve all instances of this problem in the std, as well as add a useful tool to Nim. Effectively...
Fwiw, it was always supposed to be used as "if .c file deps changed, invoke with the -f switch".
Fair enough but can't this assume e.g. "Scoop" on Windows and improve on the status quo (people not caring about Windows...)