Hacker News Reader

Comment by mantrax5 | Hacker News Reader

I'm not misrepresenting what he wrote, I'm directly responding to it.

While you may not be aware with recent language development and modern LLVM-based languages in general, to me and many others what was announced at WWDC today wasn't just a set of buzzwords and marketing speak.

Anyone who has been into language design and watched the keynote today can make a pretty good guess about the properties of Swift. There's a book with hundreds of pages of examples and description of the language semantics on iBooks right now. I've been reading it before I posted. I've been playing with the beta Xcode as well.

Swift is not revolutionary in terms of its feature set. But what makes it quite interesting is the fact this is now an official language for Apple app development, and it's quite a bit ahead of Objective-C in... yes. Safety, performance and conciseness. Fact.

View on HN

I'd prefer to see how it fares in the real world before jumping to such conclusions like you seem to be so willing to do.

You may not have been around to experience it first-hand, but we heard a lot of claims made back in the 1990s about how Java and the JVM would increase security and safety.

The arguments made then even overlap with some of those being made in this case! The lack of direct pointer access and manipulation, automated memory management and better bounds checking are some examples of the arguments used then and now.

Yet if you work with computers at all, I'm sure you'd know that many of those claims did not materialize. Flaws have been found in the various implementations of Java and its VM, and these have affected its security and safety in very serious and significant ways, many times over.

Perhaps things are different in the case of Swift. But we won't be able to say for sure until later on, once it has undergone some significant real-world use.

And you're telling me... with a straight face... that Java's lack of direct pointer access, automatic memory management and bounds checking do not make it safer and more secure compared to C and C++?

You really should stop talking.

That's obviously not what I wrote, and you would know this had you bothered to read my comment.

I'm saying that such features alone do not actually guarantee safety, if the language's implementation happens to have flaws.

There hasn't been sufficient time and opportunity to see what Swift is like in practice. It's premature to say anything conclusive about it at this point, aside from stating that we don't yet have enough information about it.

The "flaws" discovered in Java aren't in the language, but in the most popular JVM implementations, most of which - surprise - are implemented in C (for the lack of a better alternative).

Writing in Java still remains monumentally harder to fuck up in compared to writing all your code in C or C++.

Furthermore, while Java relies on said relatively complicated interpreter + JIT virtual machine for its execution, Swift has no such virtual machine. All code is analyzed statically and compiled to machine code. The Objective-C runtime which it uses (which is not new - it's the same fucking Objective-C runtime) is a tiny C library, which implements a few basic low-level routines, such as dynamic message-to-method resolution.

So next time it's best for you to shut your mouth if you're ignorant about an issue, than telling people who know better than you to "wait for conclusive evidence".

I am looking forward to see if Graal[1] replaces Hotspot in Java 9, thus replacing the amount of C/C++ code in the reference implementation.

[1] Oracle's JIT implemented in Java, from the meta-circular JVM Maxime

And quite a bit behind in flexibility and dynamism.

Giant step backwards. Vtables? I don't think so.

override keywords? WTF? Its not so much they removed the C as the "Object".

Reply on news.ycombinator.com