Posts by John Dallman 22 publicly visible posts • joined 2 Aug 2007
HP weren't getting many new sites on HP-UX Itanium at the time. Oracle thus weren't getting many new sales. If they were to reduce their efforts to "contractual obligation only", the people who'd suffer would be the Oracle customers.
They'd have to move to different hardware, and they'd likely want to switch databases at the same time, since Oracle had done this to them. The idea that they'd made a serious error by buying Itanium would not be popular with them at all.
Oracle aren't a nice bunch, but they aren't stupid, and they were right about Itanium being doomed.
AFAICS, they seem to have decided that contingency plans at intel (to drop Itanium) and HP (to port HP-UX to x86-64) that had leaked to them were the plans being acted on. Had that been the case, their actions would have been a lot easier to defend. But it wasn't. HP was determined to keep on with Itanium at the time, and it took several more years for them to see the writing on the wall.
Incidentally, don't claim that supporting Itanium can't be that difficult if you haven't tried it. It was horrid. If you needed any third-party libraries, they usually weren't available. Common programming tools were often unavailable, or worked badly. Performance sucked, and improving it was hard. The compilers were buggy. I produced Windows and HP-UX software on Itanium for a few years, and it was a constant struggle. When HP approached us to add Linux support, our reaction was "Not funny. You cannot possibly be serious, and it's been hard enough with Windows and HP-UX that jokes fall flat."
PCW was never the same after some kind of kerfuffle in the mid-Eighties when a lot of the staff got sacked. I stopped reading it after that, because the spark wasn't there for me after that. But the glory days of the early eighties, when Guy's news column really did have most of what mattered first, and the world changed several times per year... those were fine days. They were never going to last, because for software to get useful, the number of wildly different platforms had to be cut down a bit; the giddy pace of development took care of that by making most of them obsolete.
You can't cross the same river twice: I saw Jack Lang and Herman Hauser try to re-create those days in 1989 round Perihelion Ltd, the Atari Tranputer Workstation and the Active Book Company, and already it was impossible. You only get them when a new technology is exploding into usefulness, and you have to be the right age. It's happening now with Twitter and wheter the new thing is that we haven't heard of yet. It happened two years ago with FaceBook/MySpace/etc. It'll happen again: oddly enough, a bad economy isn't a big problem for this kind of thing, because it has to be cheap anyway.
But if the whole idea seems mistifying, read this story: <http://www.scifi.com/scifiction/classics/classics_archive/lafferty5/lafferty51.html>, "Slow Tuesday Night" It's from 1965, and I have no idea what gave R.A.Lafferty the idea, but it encapsulates the whole runaway maniac spirit of such times. It's only a couple of thousand words, and entirely work-safe.
John
Fans in wings hasn't been part of this project for at least a decade, if they ever were. Sure you aren't getting mixed up with the V-22 Osprey?
One reason the RN favours STOVL aircraft for carriers is that you can operate them in much rougher seas, because you can cope with much more ship motion. The USN does a lot of its training off Florida and California while the RN trains north of Scotland. STOVL landing are also easier for the pilot to carry out, so there's much less requirement for continual practice. It means you can deploy a naval squadron to do something from land, and not have to have an expensive re-certification process for carrier landings later.
If we're going to have carriers, this does look like about the best way, provided they can create a decent airborne radar system. Can we afford them? We'll, I'd cut ID cards, the NHS Spine and the super-database of all phone calls, e-mail and web sessions first, but that might not be enough.
http://en.wikipedia.org/wiki/Pentax_K_mount
Pentax DSLRs will take old K-mount lenses, and even M42 lenses with the Pentax adaptor. However, things become progressively more manual with older lenses. With newish lenses that support autofous (KAF mount, mid-nineties or later), everything works. With manual focus lenses, you get focus confirmnation from the AF system, but it can't focus for you. If the lens doesn't have an "A" setting on its aperture ring (Introduced with the KA mount in the mid-eighties), you're limited to stop-down metering.
Opinions vary widely amongst Pentax DSLR users as to the merits of using older lenses. I find it clumsy, but there are vocal proponents of working with everything manual.
The problem - in any kind of close-coupled multi-processing, you hit a point of diminishing returns as you add processors, where the coordination overheads become larger than the gains - has been obvious for quite a while. Gene Amhdal pointed it out in the early seventies. The chip manufacturers' engineers know this perfectly well, but marketing departments have been caught up in the race to the most cores, and kept on demanding them. It's just like the megahurtz wars.
They will, of course, point out that they are answering the point by putting more memory controllers onto processors - Intel's forthcoming Nahalem has three, while Sun's UltraSPARC T2 has eight - but this only moves the bottleneck elsewhere. The penalties for coordination between memory busses may become the problem, or else something else. But there isn't a Happy Hunting Ground of unrestrained parallelism out there waiting for Moore's Law to go far enough. There's only more and more struggle to make it work.
I don't expect that the manufacturers will change strategy soon; given that some kinds of code don't need much coordination, there will be example cases they can claim as "programming the right way" for quite a while. But there is no one right way, and no simple route; anyone who claims to have found it is fooling himself, usually by looking at a small set of problems, and claiming his ideas generalise to everything.
I don't know what the next fashion will be; were I given the power to decide, I'd ask either for memory that was a load faster, or some really huge on-chip caches, say half a GB. Those two ideas would have pretty similar effects: reducing the "memory wall". To get really huge amounts of processing power, we're probably going to need to climb out of the Von Neumann playpen, and deal with some different programming model entirely.
Well, it's a bit more complicated than that... Porting stuff - applications rather than OS - is what I do for a living, and I've been watching processors go by for a while, and taking notes.
People forget nowadays that Windows NT started life on the Intel i860 (Huh? went the chorus), with MIPS as its second platform, and shipped as a four-platform CD for x86, DEC Alpha, MIPS and PowerPC. And all of those platforms except x86 died. The basic reason was that x86 became the fastest, and Windows' requirements grew fast enough that using anything except the fastest became silly. XP is Windows NT version 5.1; Vista is version 6.0. Windows 9x was throughly non-portable, but it's dead.
The world is now down to just two architectures that are performance-competitive and potentially mass-market: x86 and PowerPC. IBM weren't interested in doing PowerPC laptop chips, so Apple switched to x86. That's the vast mass of Linux systems, too. Itanium is on the way out: it never really took off.
So what platforms does Linux actually have a competitive advantage on? Embedded systems, micro-portables, and the like. The issue there isn't that Windows couldn't be ported to their CPU architectures, it's that it takes too much CPU power and memory, and thereby wattage, to be a cost-effective choice. Even Windows CE, which ran on many platforms quite recently, even if it is mostly ARM now, needs a chunkier system than a stripped-down Linux, or other embedded OS. That's the killer. Not portability, as such.
They didn't just draw a million pounds from the bank and burn it. Those guys are situationists, not idiots. It was obtained from the Bank of England's stocks of worn-out cash, due to be destroyed. This cost them a substantial sum for security and insurance, but nothing like a million ponds.
It's clear that organisations who handle data about people need one policy change, straight up.
You do not put other people's personal data onto portable computers or storage devices. Period. No special cases, no approval processes; you just don't do it.
First offense is discplinary. Second is sacking. No, we don't care if you're the head of IT security. Sacking.
Bank workers don't take piles of cash home to count as part of their job, do they? The security-controlled data I work with doesn't move off its server: people know they can't have it on laptops, and they don't.
Seriously, the TSA makes a load more sense as a covert operation to reduce people's willingness to fly then it does as a security operation. Aviation is a very obvious source of CO2, although rather smaller than cars, electricity generation, and probably several other fields. The Bush Administration clearly can't say out loud that they want to shrink the aviation sector, but it's clearly a benefical side-effect of the TSA.
If you look up MS KB956139, you'll find that some of the optional bits of SQL Server 2008 won't work with Visual Studio 2008 sans service pack. They include bits of VS.2008sp1, which is due to be available Monday 11th August. Hopefully, that'll save some people a few hours of tinkering and swearing.
Up until Vista SP1, "RTM" from MS meant that software was available for dowenload from MSDN, TechNet and so on. Now, they seem to have an extra delay.
As for Balmer, this could well be positioning for a change of strategy. You know how political parties can't just change course; they have to lay the ground for saying that it isn't a change at all? MS have something of the same problem with Vista: lots of people in the MS food-chain who've been building their lives around "Vista is good" as the message for two years now. And they've taught themselves to believe that, in spite of the evidence from customers, because if they didn't believe it, they wouldn't be able to feel that they're making the world a better place. Marketing is like that, and it accounts for a lot of the strange behaviour you see from corporate marketing departments.
Setting Vista aside would mean "abandoning all the sacrifice and effort that had gone into it". People really don't like being told that they've been wasting their time, especially when their managers have been telling them its was vital. If this discussion of MS internal politics seems irrelevant, consider this: How often does Steve Balmer have a non-scripted conversation with someone who doesn't either work for him, or feel scared of upsetting him?
In our case, it's a simple lack of demand from customers. The Solaris x86 platform seems entirely credible from the technical point of view, but none of our customers seem to actually want it. Sun tell us that "loads of your customers want this", and we ask them to put the customers in touch. Nothing happens. We've been round this loop a few times now.
I suspect that Sun want the Transative stuff for (a) general flexibility, which is good, and (b) as bait to get customers onto the x86 platform. If it works, then, Sun probably hopes, the customers will ask their software suppliers to do the porting.
If the Israelis had a plausible remote hack into Syria's air defense network, then they'd only use it in a very serious situation. The chances of it working a second time are quite a bit lower. So if they did it, they must have considered that site they bombed to be the key to a getting rid of a major threat. In which case, wouldn't there have been more news about it? Overall, it seems about equally likely that somone pushed the wrong shutdown button on the ground.
Current USN jets have autoland, so so I was told on rec.aviation.military.naval. They just don't tend to trust it or use it, because you look like such a fool if you turn it on and it crashes your plane. It's usually reserved for landings with an injured pilot, or the like.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
