4 ms·
I'm going to withhold my excitement until I get proof that Apple will approve GHC compiled binaries.
by laureny 13y ago
I'm going to withhold my excitement until I get proof that Apple will approve GHC compiled binaries.
- jonsterling 13y agoA formal proof? ;-)
- shelf 13y agoA constructive proof would satisfy!
- dljsjr 13y agoWhat would prevent them? Didn't they relax the cross-compile restrictions like, 3 years ago? Adobe has a cross-compiler, that I know for sure. In fact, I did a quick Google search before submitting this reply. Relevant quote, from an Apple press release in 2010: > In particular, we are relaxing all restrictions on the development tools used to create iOS apps, as long as the resulting apps do not download any code. This should give developers the flexibility they want, while preserving the security we need.
- krichman 13y agoAt least one of Ruby Motion or Xamarin Studio is compiling source code into iPhone apps. Furthermore iPhone apps are definitely allowed to embed interpreters (caveat: no JITC's) and run scripts included with the binary or downloaded as part of an in-app purchase.
- tomasien 13y agoThey won't have a problem with it, I'd bet on it. That's not their concern really at all - the people who look at confirming apps look at human interface guidelines, functionality, crashes (although we've had an app that doesn't even launch it just crashes everytime approved), and if it meets their guidelines.
- carterschonwald 13y agoApple already approves applications written using the Xamarin tools (C#/F# compiled with LLVM), and RubyMotion. They are ok with other tools. Haskell is a more compiled language than any of F#, C# or RubyMotion (if thats even possible). There is no reason to worry about that. Worry about the normal reasons for getting an app rejected :) its also worth noting that XCode 5 (as of Dev Preview 5 or 6) has some patches backported from Clang HEAD via the Clang 3.3 Point release that fix bugs that have only ever been triggered by GHC, namely http://llvm.org/bugs/show_bug.cgi?id=16371 http://llvm.org/bugs/show_bug.cgi?id=16371 and http://llvm.org/bugs/show_bug.cgi?id=16363 http://llvm.org/bugs/show_bug.cgi?id=16363 (and I'm incredibly grateful to the engineers who work on Clang for helping get those patches backported to the pending Clang 3.3 bugfix release, and thence to apple Clang, I owe them all a beer or something like that) I think Apple, as a business, wants every community of smart Developers to write amazing, profitable applications for their Desktop and Mobile ecosystems. Haskell targeting iOS immediately grows the pool of smart developers who might write iOS applications, and hopefully also just make writing certain very interesting applications much less painful! This is exciting! tl;dr the latest xcode 5 Apple Clang dev previews have backported patches from Clang HEAD fixing bugs only ever triggered by GHC. Interpret that however you please :)
- tluyben2 13y agoThere now are many interpreted as well as compiled non Objective-C languages around; from .NET languages to Object Pascal, Ruby, Java, Lua, Python, Javascript and others used to create iOS applications and they all are happily allowed in the Appstore. You are just not allowed to dynamically load 'new code' from servers and execute it on the device unless it's via the Safari Javascript engine. Compiling to ARM and then running on the device is perfectly fine with Apple.
- aaronbrethorst 13y agoYou would be amazed by how many top-tier iOS games were written in C#.
- Nekorosu 13y agoThat's not an issue for a long time already.