Posts by Jasmine Strong 91 publicly visible posts • joined 13 Feb 2007
how exactly do you plan to test its performance on real networks? It's not like AT&T works indoors, especially in big ferroconcrete buildings like 1 Infinite Loop. This isn't Europe where mobile networks actually have link budget; this is the USA, where cellcos suck.
Polaroid's new owners are making a new camera to work with the film, and the Impossible Project guys are working on one too. There are also plenty of deadstock brand-new cameras out there- you can even get them on Amazon. It's not like they'll stop working in the box, especially since there's no battery in the camera.
...more than half a million packs of Polaroid film that they bought up in the original liquidation, over the last year and a half, it seems like there is a market- just not the one you seem to think.
My experience with this is that Polaroid is an excellent niche market, just not the mass-market product you seem to assume. When you're at a party and you offer someone a polaroid snap, you don't just take a picture, you also give them a gift. There's a unique tactile element, too, that you don't get from digital photos, where everyone clusters around to watch the photo develop. There's a magic in watching it- and people seem to respond to it.
If you've got an old Polaroid hanging around, and you attend gatherings, you could do worse than to blow a tenner on a pack of film for it and see.
Then there are a number of factors that really make Polaroid irreplaceable- firstly, that lack of a negative means that once the photo's developed and set, it's set in stone. This is something digital formats just can't replicate, and it's something that gives polaroid photos more weight in court cases. This is very handy when you want to sue that builder who screwed up your house.
Secondly, the film is huge compared to a 35mm negative- easily the largest consumer format- which means you can actually blow them up really large before they lose detail, at least if you're using an SX70 or Spectra camera that has a decent glass lens. These cameras are far more compact than other large-format film cameras, and a lot cheaper too (you can pick them up for a few quid on eBay.)
Thirdly, if you're at a party and you take a picture of someone, and they don't like it (or they're doing something that might get them in trouble...) then you give them the photo and they know for a fact that you don't have a copy to upload to the internets. That's a big win in my experience.
I ordered a Dell Mini 9 from them many months ago- it took over two months for them to actually ship it. Apple, Lenovo and HP are making basically the same product and they can ship machines the same day you order them, especially with a one-size-fits-all SKU like the Mini 9. What is wrong at Dell? Surely they're not that popular.
In fact, the ability to run Single-Frequency Networks was one of the design criteria for DVB-T. The UK runs a Multi-Frequency Network because it was first to implement, but many countries are on SFN. The bandwidth savings are substantial.
The way this works is that each symbol in the signal has a "guard time" around it to allow for the "ghosts". It works really rather well. It doesn't fit nearly as well in the substantially less sophisticated 8VSB modulation used in the ATSC system, but with enough DSP at the receiver it can be made to work; indeed, for digital TV to work in cities with their multipath-intensive propagation environment it *must* be made to work or most people will not have service.
This is pretty easy. You have a GPS tracker already, so just use a nylon tether between the aircraft and the balloon and have the GPS controller connect the battery to a bit of nichrome wire wrapped around the tether. When the GPS shows that the aircraft is falling, turn off the current. Shouldn't weigh more than a few grammes.
Because your router doesn't have to handle lightning, homeless people, vandalism, floods, hot summers, freezing conditions, the occasional car crash, or, oh, fifty subscribers worth of pirate movies and pornography.
Just you, which only counts as "vandalism" and maybe two or three normal people's allowance of porno.
Neutrons, that is. Thermal neutrons will be produced in large measure by these reactors, whether they're dosed with Am-241 or not, and that means that suitable clandestine addition of natural uranium around the core can be used to produce Pu-239. Which is what they want.
The key to making spent fuel impractical to reprocess for weapons use is to mandate long fuel cycles, which load up the fuel with undesirable Pu-240, which is both very hard to separate from Pu-239 and highly radioactive (worse than the usual stuff, anyway, which is pretty unpleasant by all accounts.) None of this stops a suitably conniving reactor operator from sticking extra stuff in the reactor to irradiate it with stray neutrons- the only way you could do that would be by absorbing them all, which would, uh, stop it working.
I guess you could probably design a reactor which didn't have any space to add anything, but then where would the coolant and control rods or drums go?
The other point I'd like to make is that now laser isotope separation is possible, it seems unlikely that anyone with enough money and sufficient insanity to actually want nuclear weapons is going to be stopped by any technological means. Pandora's box is well and truly open, and any half decent A-level physics student could design you a pretty reasonable bomb given access to the interwebs and a couple of months to do the calculations involved.
What exactly are you getting at? XScale is not a modern processor. It's an ancient, shonky rehash of ARMv5. There are ARMs that are more than ten times as fast as that now.
As for the matter at hand, it seems ridiculous given that they discontinued the damned thing. They might have had a point if they'd updated it.
Actually, it has nothing to do with the line timebase or CRTs. The way they used to work was to pick up the local oscillator- this is part of the receiver that picks up the video signal, and even LCD TVs have it (only the direct TRF type of receiver, which are quite rare in TV-land due to poor sensitivity at that sort of bandwidth, don't have them).
The local oscillator generates a signal a few MHz below the signal you want to pick up. This signal is mixed with the received signal, and the difference between the two- the beat frequency- is the signal you want, shifted down to a lower frequency. This is called the intermediate frequency signal. That is then demodulated and fed to the rest of the TV receiver.
All of these signals are generated by the TV, and all of them leak out through the incoming aerial cable. There are filters to reduce this, but since they're analogue technology they're not 100% effective and a sensitive receiver can pick them up.
However, the way they do it these days involves a database of every address in the country and the SQL statement "SELECT offender WHERE licence EQU 'no'". After all, what's the point of a detector van when everyone has a telly? The detector vans you see driving around these days are fake, and intended purely for scare purposes.
The lunar surface is a good place to put unshielded radioactive things. After all, the place is constantly deluged with radiation from the Sun anyway.
If the radiation worries you that much, just dig a deep hole and bury the thing. It's not like there's any risk of groundwater contamination.
Disclosure: I used to work for Symbian, in the kernel group
It seems unlikely that most of you naysayers have actually seen inside the belly of the beast. The core of the thing's actually really good. It does some stuff that other OSes can't even dream about, like running hard realtime with untrusted user apps and that sort of thing, and it has pretty exceptional hardware support.
What cripples Symbian is three-fold: poor management, terrible processes, and a choice of weird UI layers. No doubt half the senior staff are going to leave over this open-source business, so that ought to clear up at least the first problem.
What I'd like to see is a decent UI/application layer ported onto Symbian's unquestionably good underpinnings. S60 and UIQ are the problem with SymbianOS, not the actual OS itself. Symbian's increasing moves towards POSIX compliance may make this more likely to happen...
You said:
BTW for a system that allows users near-unlimited choice of apps, a chip with virtual memory management is essential. ARM doesn't have it. So however good it is for phones and suchlike, it'll never break into the general-purpose small computer market.
Uh... ARM has had virtual memory management since ARM1/MEMC1 days. ARM has *always* had a memory management unit. The only ARMs that don't have one are the ones that have had it taken out (for example, the tiny ARM7s that control toasters, or the ARM946 in a Nintendo DS).
ARMv6 (ie., ARM1136 and all those processors now going into cellphones and N810s and such) and later even have physically tagged cache, so they don't waste memory bandwidth running OSes that map all applications to the same address (like Linux, for example.)
So, uh, you're completely wrong mate. Sorry.
The winged orbiter may be sexy, but it doesn't really work very well for what you want it to do. Think about it: on every launch you lift fifty tons of delta winged space bomber into orbit, and you bring it right back again. Why would you do that? Just launch the bits that need to stay up there separately, and stick the people in a minimum spacecraft capsule type thingy (X-33 would have been awesome, for example).
Poor old Shuttle, all stretched out of shape by its conflicting design demands, ever-shrinking design, horrible reliability problems, retrofitted safety systems (two separate redesigns and still huge windows of the launch where survival is impossible in any kind of major failure) and antique electrical system. It could have been rather beautiful as originally envisaged, but it turns out, things just can't work out that way. There's a reason the Russians only flew Buran once. It's because they hadn't killed off the Soyuz programme ahead of Buran becoming available.
There's a reason it has the highest suicide rate in the world, and that's because living in a closed society with only horizontal movement is very depressing.
Switzerland, despite its low taxation, is a very unpleasant place to live. Trust me, I spent more than a year there. Everything that is not mandatory is forbidden. Society is controlled with an iron fist, conformity is king.
And as for the great standard of living- that only really exists if you ignore the vast population of stateless people living in Switzerland. The undesirables, the foreigners and the unusuals, trapped but unable to leave. These people have hard lives because Switzerland does not offer them a path to integration.
It is a staggeringly unjust society. Unless you have seen it first-hand, you can't believe that something like this exists in western Europe in these post-Iron Curtain days.
...that all you need to support cellular data is a SIM slot.
Actually you need lots of things:
i) a SIM slot
ii) a bunch of radios (at least the GSM ones, probably UMTS ones too, and probably on a few frequencies, too)
iii) some antennae (see ii)
iv) if you're running Linux, another processor to run the signalling
v) A signalling stack (very expensive to licence and implement)
vi) regulatory approval from every country you're going to sell the thing in
This is going to cost you a lot. It will substantially increase your BoM and development costs on something that you may only sell a few thousand of. Are you going to bother? When everyone already has a phone anyway? Of course you're bloody not.
These systems and their associated spectrum will only become available after the shutdown of analogue TV anyway. DVB-T is far less sensitive to interference, at least if the potentially-interfering signal is designed appropriately. There's no problem with "what if they're not transmitting when you check" in the TV bands (roughly, bands III, IV and V, depending on where you are) since TV transmitters are modulated continuously.
It certainly *does* use wear levelling and bad block management (they are mandatory for NAND devices, since not all blocks are good from the factory). However there is no getting around the fact that writes to NAND are slow (in comparison to writes to RAM) and power hungry (consuming thousands of times more energy than reading).
I haven't spotted your power monitor comments- I will check them out.
Firstly, SymbianOS has supported "virtual memory" in the sense of "a paged environment in which the operating system uses a memory management unit to form noncontiguous regions of addressible memory from one or more contiguous pieces of physical memory" since the very beginning. The new EKA2 kernel architecture present since SymbianOS 8.1b has extended this, with support for pluggable memory models and new, more sophisticated process isolation (to allow better security for wireless signalling stacks).
Secondly, "demand paging" means that application binaries are only loaded from backing store (in this case, Flash memory) into RAM when they need to be, reducing application load times, helping the RAM controller keep physical pages closed (in retention mode or even switched off!), and making things much more efficient at the expense of some compromises in realtime performance and fairly wide-ranging complication to the message-passing architecture of the kernel. Most SymbianOS devices load entire binaries into RAM at a time, which makes more sense when you understand that SymbianOS is an intrinsically object-based environment and code objects were not expected to get large, ie., each app was expected to comprise many small binaries. Large facilities such as the UI layer would run directly from NOR (parallel) Flash memory, and not have to be loaded. Things didn't quite work out that way. NAND Flash is far denser and generally faster than NOR, but it has semantics much more like a disc than like a memory array, and so binaries have to be loaded from it and held in RAM- including the OS.
Thirdly, the new features in N95's memory architecture are certainly supported by the kernel- indeed, they are substantially implemented in the OS nanokernel. SymbianOS may be intended for resource-constrained platforms, but it is far from feature-poor.
Finally, the real reason that there isn't page swapping is that writing to Flash is slow, power-hungry, and wears it out. You don't want to write to Flash too often. Swapping to microdrives has been discussed a few times, and they don't wear out so fast, but until someone comes up with a way to make phone batteries last a lot longer, it's still not really an option.
Hope this clears up some misconceptions.
...because my N95 has plenty of RAM to play videos at the same time as it runs the energy monitor; not a lot of point in having one otherwise, is there?
It uses about three watts. However, note that:
a) the energy monitor itself burns about a third of a watt (which is an order of magnitude more than the normal quiescent power of the N95 as measured with my handy dandy milliwatt meter.
b) running two apps at once is disproportionately hard on the battery in these devices due to the cache pollution side effects; ARM1136JF-S doesn't have that much cache, and all code segments are cacheable in SymbianOS, so the RAM interfaces will be working much harder with the energy monitor running than it would be otherwise.
Measured figures with a meter:
Typical power dissipation of an N95-3 playing AAC audio at 320 kbits/sec into 32 ohm headphones at maximum volume is around 300 mW with display illuminated and 200 mW with it dark, while on a rather weak UMTS HSDPA cell (-= 30 mW for GSM/Edge), Bluetooth on (probably <10 mW- was lost in the noise), WLAN off (depends on your power setting, but generally double the setting's ERP), and with no USB connected (+= 100 mW or so for the USB line drivers).
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
