Re: Phoenix language

[Ivan Vučica]

On Wed, Oct 22, 2014 at 5:23 PM, Dr. H. Nikolaus Schaller <hns@computer.org> wrote:

I see.

Shouldn’t they have much code in common (except a different grammar)?

I don't know enough about Swift or compilers, but I would not expect that to be the case.

From what little time I spent looking at Swift, it seems to be sufficiently different from Objective-C to warrant a different approach. It may be able to interact with Objective-C classes and objects, but it seems to be a sufficiently different language.

 

Ah, nice to know.

How is that related to the GNUstep work? Does it help GNUstep or does it distract Greg

from working for GNUstep?

Obviously only Gregory can comment on that, but from my (very small amount of) chatter with him, I think this project is being undertaken to future-proof GNUstep in case it turns out people greatly prefer Swift.

It may also turn out that Swift really is a nicer language. Even if not, I appreciate the ability to quickly put together prototypes.

But since for the past year I have written very, very miniscule amounts of code targeting STEP-like environments, I have not found it worth my time to learn a language that, to me, innately seems less flexible. So I'm a wrong person to continue commenting on whether Swift is a distraction or the future.

John Siracusa's review of Yosemite is extensive; for this thread, the embedded extensive overview of Swift is more interesting. I would highly recommend reading pages 21, 22, 23 as an overview of what makes Swift interesting. It sounds like the added step of compiling code into SIL before compiling and optimizing SIL into LLVM IR makes some interesting optimizations possible.

  http://arstechnica.com/apple/2014/10/os-x-10-10/21/

I'd be particularly interested in hearing from Gregory what is the intended pipeline in Phoenix and how it compares to what Swift compiler is doing. 

And, of course, I'm interested in what our resident compiler experts think about the architecture that Siracusa has described.

--

Ivan Vučica

ivan@vucica.net