While I'm concerned that Chrome is the only browser supporting native client (Mozilla outright rejected it), it's still awesome to see lua in the browser.
Besides IE, Opera, etc all rejected it as well.
The fact that this time it is from Google doesn't make it better.
We slowly rid ourselves from stuff like Flash and Java on web pages and now we want to make the same mistakes again?
I know, it isn't bad in first place (Flash, Java, Active X weren't either), but I'm pretty sure we'll see better approaches.
NaCl is a platform independent secure code execution environment. It is NOT a way to talk to the underlying OS like ActiveX. Is is NOT a way to write plugins like ActiveX.
It's effectively just another way to write code in a browser like JavaScript. Is has no more or less permissions than JavaScript. It is no more or less access to APIs then JavaScript. It just has 5 to 100x more speed than JavaScript depending on what you are trying to achieve.
Do you have any sources explaining Mozilla's decision on this? I'm curious as to why they would reject it...
I think that they have some private issues regarding a plugin controlled solely by their browser competitor who wants NaCl apps to be purchased through the Chrome Webstore.
PNaCl makes things a little better by using LLVM IR instead of machine code, but one of the LLVM developers has publicly stated that "LLVM IR is a poor system for building a Platform" (http://lists.cs.uiuc.edu/pipermail/llvmdev/2011-October/0437...).
I don't agree with this. From what I've read NaCl isn't even getting auto-run by default. It is not trying to replace JS just sit alongside it as an application language. It lets developers target CPU architectures instead of Operating Systems which makes a lot of sense to me.
Also it's C/C++, any popular app would be easily ported to mobile devices.
Look how far JS has come in 5 years, imagine how for NaCl could have come.
On your PNaCl point, it may not be optimal but I love the idea. "Compile to javascript" is an awful concept, "compile to IR" is better for a number of reasons that I won't go into right now.
Even if it's not trying to replace JS, if it works well enough, a lot of sites will use it to replace JS, whether because they want to add functionality that's not as easy to implement in JS, because they don't want to expose their source, or just because their programmers don't know/like JS. We have already been through this with ActiveX, which was a similar idea, but poorly executed.
> Also it's C/C++, any popular app would be easily ported to mobile devices.
Only if the developers care. There's a chicken-and-egg problem here. If your mobile platform can't interact properly with a large proportion of the web, no one is going to buy it, and so no one will bother to port their site to it.
> "Compile to javascript" is an awful concept, "compile to IR" is better for a number of reasons that I won't go into right now.
Possibly. If you absolutely have to run C/C++ app on the client side, I agree that something like PNaCl (but actually architecture-independent) is a better solution than something like emscripten. But I think it would be even better to 1) run it server side or 2) write it in JavaScript.
tldr: Nacl cannot be standardized, and the relevant API's are too strongly tied to chrome/webkit details to provide a first-class port to Mozilla.