I agree with you that MS Word is not a programming language. However, it seems to me that you are conflating the concept of a general purpose programming language with that of a programming environment that BAG is talking about. He specifically refers to "domain-relevant expression tools" as a description of the programming environment.

Expressing a domain concept can be as simple as drawing figures on the screen *for a drawing program*. A drawing DSL or a robotic control DSL, for example, is not intended to be used for writing "Hello, world" on the screen!

I think we all can agree that you should use the right tool for the task at hand. In this context, that means choosing the right level of abstraction. Trying to write a Facebook application in Assembly language is just silly. Is Facebook a general purpose programming language? Of course now. However, it certainly does provide a programming environment.

<quote><p>Now go try that again using an EDGE or CDMA network communications device.</p><p>Also, go try doing that basically with any cellphone-like device, which is what this is aimed for. You won't get the same response times, I assure you.</p><p>Remember, this OS is aimed at machines that load everything into RAM and cache to a micro-SD card or something, tops. No hard drives allowed, and no super fat-bandwidth pipe.</p><p>You'll be lucky to get the performance equivalent of a Pentium 3 sending data over a 56k modem, on average (cell provider's claimed network speed is rarely if ever actually realized by any of their customers in any location, burst speeds at best, and the really depends on signal interference and network traffic).</p></quote>

You obviously never even tried this. On a "cellphone-like device", the mobile interface is actually very fast. On a droid, Gmail is every bit as quick as on my laptop.

You know... it is possible that the Google algorithm was tweaked between the time he did his search and the time you did yours. I'm not defending the guy, he might be full of shit or not. I'm just saying it's a possibility, right?

What you say might have some basis in reality but I have to say it sounds like bullshit to me. Java could easily have provided for manual memory management without using direct pointers to memory in order to do so. Hypothetical allocation / deallocation methods could simply operate on references that may only have either a legal value or null (when deallocated or going out of scope on stack allocation).

If you bothered to do the slightest bit of research, Gosling himself says many times that it's for programmer convenience and to increase the robustness of applications. Pointer bugs are the bane of C and C++ programmers (yes, I've programmed in both for about 15 years and had my share of them as well). Gosling was shooting for something like C++ with all the hard bits removed.

Can you provide a reference for your claim that it has nothing to do with programmer convenience and is only about the Java security model?