You're going to be in for a rude surprise.

OpenGL drivers on Windows are awful, DirectX is where all of the development effort goes on driver teams. At work we wrote our app using OpenGL for a 3D overlay because we ship on Windows, Mac, and Linux, but on Windows we took the time to write a DirectX backend instead of OpenGL and the stability and performance shot up noticably. OpenGL is a forgotten "checkbox feature" on Windows today, not much more.

The Xoom was half-baked and lacklustre, and no other tablet has been widely available for a reasonable amount of time.

That's all there is to it.

multi-touch is not a unique property of iPhone OS. My Nexus One is multi-touch, as is the Palm Pre/Pixi and so will be Windows Phone 7 devices.

Why not use system codecs? Because you're not solving the original problem. The problem was dependency on externally variable code that was frequently proprietary to render the web. Instead of requiring Flash to view websites in a default web browser, now we'll require K-Lite Codec Pack Max Extreme 2.0++?

Which codec would you use? Theora isn't implemented on most systems. h264 won't work on Windows by default (7 will, but that's just a small portion of the market). VC-1 won't work on Mac and Linux. MPEG-2 isn't even guaranteed to be found.

Congratulations...you've just opened a whole 'nother can of worms.

That's not true at all. Firefox, since 2.0, has had memory caching just like Opera does. Go to about:config and tweak how much is used (which is default-set based on your system capacity, just like Opera). It's called "browser.cache.memory.capacity".

I understand why they don't agree with the practice, but this is hardly front-page Slashdot news. The summary is, if anything, very misleading. This has NOTHING to do with Mozilla's plans for HTML5 or web openness, it's everything to do with some nameless blogger disagreeing with another nameless blogger's implementation of video fallback.

This is non-news, to say the least.

Some random Mozilla Hacks (note the word Hacks) blogger posts some code that web developers can use to implement HTML5 video (which does not use javascript, contrary to the implications in this article and summary?) and also provide a fallback path for non-HTML5 Video browsers (IE, Opera, etc). Their particular method of providing the fallback code uses javascript to determine browser capability, and uses Flash if HTML5 Video is not there.

Why is this upsetting to anyone? The implication from the summary is this is a less "open" way to do it, but last I checked Javascript/ECMAScript is a standard that all browsers implement already.

I cannot fathom why anyone would be so upset by some blogger providing JS-implemented video fallback implementations.

Firefox 3.5 implements the majority of those. IIRC Webkit and Safari are almost there too, and even MS is making a sprint to HTML 5 -- IE8 has started implementation there.

As for editors, yes -- they are under development.

Apple was far from the first company to implement multi-touch, it's just the first people think of. Multi-touch, and its patents, go back for quite a long time.

By definition, a superscalar processor does not need to be instruction re-ordering, speculative execution, or branch prediction. All a superscalar processor needs to do to be called superscalar is to dispatch more than one instruction per clock cycle to redundant functional units on the processor -- the 360's CPU is absolutely superscalar.

Incidentally, you are incorrect other aspects anyway: the Xenon cores DO have branch prediction, just in a significantly diminished capacity compared to what we're used to on the PC (small history tables, simpler logic, etc). The Cell's SPEs have none at all, and rely on "branch hints" instead where the programmer or compiler specifies which path may be likely at any given point. As a result, it also has speculative execution.

But you are correct in that the 360's CPUs -- like the Cell's PPU and SPEs -- is in-order and thus does not re-order instructions. It is up to the compilers to generate optimally ordered instructions, which isn't as big a deal as you may think on a closed-box system with optimized compilers. Back in '03-'04 I worked in IBM's compiler group on this very thing.