search

lambdacube3d / lambdacube-compiler

LambdaCube 3D is a Haskell-like purely functional language for GPU. Try it out:
http://lambdacube3d.com
Other
85 stars 9 forks source link

An attempt to fix https://github.com/lambdacube3d/lambdacube-compiler/issues/11 #12

Closed deepfire closed 6 years ago

deepfire commented 7 years ago

This seems to work for my test application.

EDIT: there was a controversy, due to launching a different application and thinking that something is broken.

csabahruska commented 7 years ago

What version of megaparsec do you use?

deepfire commented 7 years ago

The conflicting instance is introduced in upstream commit https://github.com/mrkkrp/megaparsec/commit/dd2386aafc6b23553dda3ae0bc97cba80cf4124a#diff-7762a64005e439e93f28d5c9a983baa9

csabahruska commented 7 years ago

sure but there is a megaparsec <5.2 constraint in the cabal file

deepfire commented 7 years ago

Sigh, yes, but.. I had to "jailbreak", using Nix terms, because I had to go with the latest nixpkgs.. Of course, with the benefit of the hindsight I would have overridden megaparsec to an earlier version -- but that wouldn't have been free either, since it would've meant rebuilding things from source..

It's not a simple picture, and I agree, personal inclinations did play a role in this.

But in any case, we now have an option to have compat with the latest megaparsec : -)

csabahruska commented 7 years ago

I see. I'm not a nix user. Usually I use stack or cabal new-build in local sandboxed mode.

csabahruska commented 6 years ago

Why did you close the issue? Did you solve it?

deepfire commented 6 years ago

The latest lambdacube-compiler builds with MegaParsec 6.3.0, with #13 -- so I've closed that issue.

Also, the changes from #13 solve different MegaParsec issues -- the RWST instance conflict apparently is no longer a problem.