@TravisMWhitaker@angerman some may but requires extra work, the llvm one i handed over patch to miso author to proceed; some only makes sense for ghc (e.g. the ghc shared library has too many exported symbols that chokes v8 default limits in nodejs, similar issue to win32 dll)
@angerman@TravisMWhitaker fwiw development does happen on the master branch and is subject to ghc contribution standards. it's just the release branches that i fork and cherry-pick the new features & bugfixes, cause i don't want to wait for months for good stuff to reach end users
@FlyingGuibou @mattoflambda it makes loading code from home modules faster, but not across packages, for those ghc still loads shared libraries. still definitely worth a try
@kerckhove_ts use alpine repo's ghc and only pass -optl-static to the final executable target, official/ghcup alpine3_12 bindists don't work well. or use musl64 cross target of haskell.nix
@bgamari@per_arneng it complicates the build story a bit when targeting browseres, and we also need to be able to generate raw haskell bindings from .wit interfaces. so i'm prioritizing on other stuff atm. we'll definitely support wasm components at some point, rest assured
@dr_c0d3@tweagio a large number of tests get skipped is expected behavior of the testsuite; e.g. if you look at a recent fedora33 job at https://t.co/QKfkAfuVqZ, 33117 out of 43205 cases are skipped. the real meaningful numbers are expected/unexpected passes/failures.
@KevinHoffman@tweagio Currently foreign c exports should work, both static/dynamic, but it takes some exotic linker flags to get this done. Reping me (preferrably not on twitter) if I haven't posted a gist example a few days later..
@programmer_just@tlively52 To be clear, wasm gc definitely has a huge value! But cross-compiling existing language runtimes may pave a smoother path to wasm adoption, especially when garbage collection is tightly coupled to other conventions in the language
@programmer_just@tlively52 We did look into the possibility of using wasm gc. Other than the amount of work needed compared to leverage the existing one, there's the lack of weak pointers and finalizers in wasm gc mvp, which is a problem for Haskell given weak pointers are actually being used a lot.
Huge news for Haskell today: our WASM backend has been merged into GHC! Compared to our older "Asterius" work, the result is greater speed and support of advanced runtime features. ๐ https://t.co/mHg8KtKjFN