jrabone

Other crows might - apparently they can remember faces and summon other crows to mob offenders on sight...

http://www.sefs.washington.edu/research.acl/Crows_and_Other_...

Hmm. Would you buy phone components from people who can't do Unicode? "It�s pretty cool." Revenge of the SmartQuote.

Just need to find a CL developer for hire...

£800 option on the F30 in the UK - doesn't seem that much on a £43K 330d...

I think there's too much going on in that HUD for the average driver to parse. I'm also unconvinced by the choice of elevation perspective for the route guidance - even though the display is above the road, it feels less confusing to have a "birds-eye" perspective on the route guidance.

As the Sheppey pile-up demonstrates, there's a lot of people who simply suck at driving - specifically maintaining sufficient braking distance and NOT cutting into someone else's braking distance when changing lanes (in poor conditions that space is for ME to stop in, not for YOU to slide into). As it is, I'm conscious that I need to be aware of the car behind as well as the car in front. The only way it'll work is if the car brakes for you, and then everyone has to have that technology.

Whether to provide BCD optimisation always seemed to be a tricky engineering decision; virtually nobody used the 6502 BCD instructions in the amateur home microcomputer environment I was familiar with in the 80s, but it was clearly considered to be important to the CPU manufacturers. Were there BCD benchmarks back then? Was it considered a killer feature to make financial software easier to write? Did Rockwell ever capitalise on that patent?

...thus pissing off everyone using Hangouts for tech support to people who are only ever going to have Internet Explorer installed. Thanks a bunch, Google.

I really want to crack some skulls together over the state of video compression codecs and OS support. Same thing happened with all the circle-jerking about HTML5 video - Flash had AS STANDARD a bunch of useful codecs (eg. Screen Video) for which there is no HTML5 good replacement if you want to do a lossless screencast.

Interesting that the power plate indicates 60 Hz supply - yet the UK is 50 Hz. Perhaps stretching things for the title to claim it's "from" the BBC, as opposed to "as used by" - this one was most likely was made for the US market. I wonder what its provenance was?

Good for you. If you hit the bug, it'll look like http://forums.androidcentral.com/attachments/google-nexus-4/...

I'm happy that you haven't had this problem (although without a screen I'm not surprised you get multiple days use from a single charge.) Unfortunately, lots of people are not so fortunate, and I'm one of them. There are a number of known, reported, acknowledged bugs that might or might not be fixed in the latest patch update to 4.3 that cause "Android OS" (actually a partial wakelock held in the kernel) to stop the phone from going into deep sleep and draining the battery. For example, https://code.google.com/p/android/issues/detail?id=52034 is one such bug. There are others (I don't use a wireless charger).

Until whatever byzantine download scheduler deigns to push the patch to 4.3 to my phone, it's basically unusable.

Yeah, you too can watch as "Android OS" drains the battery in 6 hours. Honestly, if you get a Nexus 4 be prepared to be a beta tester, or scratch monkey every time there's an OTA update. Perhaps it's just "international" users, but I totally regret buying into the Nexus hype.

And then there's the hardware issues (bluetooth kills WiFi if you try to use both at once), firmware issues (WiFi doesn't roam properly, ARP offload broken), straight-up bugs (wireless charging power management holds permanent wakelock preventing deep sleep)...

http://forum.xda-developers.com/showthread.php?t=2072930 really shows what Android is capable of.

Some of these algorithms (eg. Knuth Shuffle) are implemented using Java's Random, as is their StdRandom helper class. While they do state their assumptions, I think it's worth making the point that java.util.Random has traditionally been a pretty poor RNG (it was a simple 48 bit linear congruential generator with some appallingly non-random behaviour in the least significant bits). It might be better in Java 7 - I haven't looked. Anyone doing serious work with RNGs eg. for Monte Carlo sims would be advised to go elsewhere. I'm not talking about secure RNGs - that's a different ball game. Anyone doing crypto with java.util.Random has gone FAR beyond von Neumanns 'state of sin' ...

But instead required a bunch of moderately obscure changes to system config which you are bound to forget after a few years when you reinstall / image a new machine. The Virgin Superhub is a crappy barely-consumer grade box from Netgear with firmware written by Virgin. Modem mode is all it's good for. Sometimes the right answer is to spend the money on a decent router - Draytek are passable.

You can probably do this with love hearts or similar - buy a couple and reassemble two tubes as matched, randomly chosen one time pads. Entropy would be terrible though.

How is that different to the web of trust between friends that the parent referred to?

You need a CA for TLS-secured comms between MTAs. Many MTAs are set up to do opportunistic encryption out of the box, but they won't be validating the certs they get, so there's no guarantees about who's got the private key.

Of course there's no need to rely on a central CA - it's not hard to run your own, and you can make it reasonably secure - say, a small ARM Linux board with passphrase-protected private keys on a removable SD card. Generating the keys in a secure way is perhaps a bigger problem - untrusted hardware RNGs, poor quality entropy after boot etc.

One can also keep a virtual machine CA on a hardware-encrypted USB device (IronKey, say, depending on your level of trust with Imation) - it's easy enough to bootstrap a tiny Linux distribuution from source with just OpenSSL and some utility scripts to issue & revoke certs.

You don't necessarily need a $5000 HSM solution to issue your own SSL certificates at this level.

Indeed; and the bit-packing code to convert 8 7-bit characters to 7 octets is a PITA (I wrote a Java SMS server)

I expect that applies equally well to the UK with gantries of ANPR cameras over major motorways as well as portable ones in laybys, like the one I saw this morning on the A1 leaving Edinburgh. It wasn't a police car, though, but a dedicated "road tax enforcement" vehicle running it.

I was reminded by a colleague today of a quote attributed to Roger Needham, something along the lines of "privacy is a transient condition in between people realising there isn't an omniscient god, and government realising there was a vacancy" (quoting second hand from memory)

Build it from audited source?

As for updates, I imagine if you set up a domain you can run your own WSUS update server, MITM the connection, etc. - and then compare the behaviour with a "regular" home PC.

The problem really is how deep the hole goes - as per Ken Thompson "Reflections on Trusting Trust", 1984.

Not in my experience. I'm using greylistd with Exim and a custom config which uses geolocation to do greylisting by subnet country code. I've only needed to whitelist a handful of broken MTAs which don't retry.

My spammers clearly haven't caught on yet...

      zgrep grey /var/log/exim4/rejectlog* | rev | cut -c 2-3 | rev | sort | uniq -c | sort -rn | head
      8 CA
      7 IT
      5 ES
      4 BR
      4 AR
      2 TR
      2 RO
      2 IR
      2 EC
      1 PT

But WHAT, exactly, can't we trust? I've seen NO technical detail to any of these discussions, yet there are a number of sub-systems that might be compromised:

- low-level crypto APIs (the 'DLLs' referred to obliquely in the article); these are more interesting. I imagine they could be compromised for weak session key generation or other leakage of key / plaintext, or generate the session key in such a way that the mythical 'NSAKEY' can decrypt it. Huge impact, if so, but only to certain software; AFAIK Mozilla doesn't use the Windows crypto API / certificate key store (but Chrome does).

- SSL certificate generation (built-in CA for Windows Server builds); certificates stored and replicated via Active Directory; does anyone actually use this? In fact, does anyone actually use client SSL? It is likely also used for domain peer replication, which could potentially be over an external network (but why would you not use a VPN there?)

- Encrypted File System; already contains an escrow key-recovery mechanism to allow administrators (including domain admins) to recover a lost user key. Only likely to be relevant if hard disk or backup images seized, so less impact.

- BitLocker drive encryption; similar to EFS but uses a hardware TPM and is per-machine rather than per-user. Fairly sure escrow key recovery at the domain level is possible here too. Again, only likely to be relevant if hardware or backups seized.

- Office document encryption; did anyone SERIOUSLY think this was worth using anyway? There are so many key recovery services out there for this (Elcomsoft et al)

- Communications applications (Skype et al); again, did anyone SERIOUSLY think this wasn't already being monitored, even before Skype became a Microsoft product?

- Some other OS-level 'phoning-home' behaviour. I simply don't believe that no-one has spotted this happening, if it's there - we can do traffic analysis too, and there are plenty of people running Wireshark on their own networks.

"All you need to do is to send emails to us." - no thanks. If you've built something more sophisticated than I can implement with Perl and Exim's system filter functionality, I might be interested in giving you money to save me my time. Maybe a per-domain license?

Not sure why you got downvoted for this, other than brand snobbery. That's also been my experience with Shure. As well as Sennheiser HD650 & Fostex T20RP, I own pairs of Shure e5c and e215 in-ear phones. The e5c are getting on a bit, and are perhaps fragile, but I had custom ear molds made for them. Out of the lot, the e215 are the best bang-for-buck I've come across.

The e5c's main problem is they are SO sensitive they are completely unforgiving about the amplifier used - almost anything with a gain block after a volume control is going to have too high an intrinsic noise floor for them to not show it up.

When my Amazon colleagues first introduced a carousel implementation as an optional widget about 5 years ago, I believe they were wildly successful in A/B testing for a period of time; then we went through every page having about 6 of the damned things, then they basically disappeared... With enough users, sometimes A/B tests can degrade to novelty detectors.

Sounds like the sort of problem the Mersenne Twister PRNG is good for.

ABP has had an option to enable some white-listed ads for a while. It's ticked by default, but it's not exactly hidden...

If you need more power you have 2 options. You can add more batteries in parallel and increase the size of the cables to take the higher current, or you can go to a 24 volt system which means you can use the same cable. 24 volt components are maybe harder to find though. My system is really intended for short periods of peak power (say 30 minutes a week running the mower). It was harder to find cheap inverters to handle more than 1kW at the time too.

It was an Elecsol 100Ah leisure battery. The equivalent would be http://midsummerenergy.co.uk/buy/Sealed-Batteries/PM-110.htm...

Other bits were a Morningstar SunSaver solar regulator SS10 and a Nasa Clipper BM1 Battery Monitor (unnecessary extravagence - don't bother!). I also added a couple of 12V fluorescent lights powered from the SS10 so I didn't need the inverter running just for light.

It's a pretty simple set up, the hardest part is making up the high current battery cables - the clamps, fuse holder and connector to the inverter are pretty substantial to handle 100A.

I mounted the panel on the roof of the Apex shed with simple aluminium brackets. The wiring is pretty simple (more complicated by the battery monitor which has a 100A shunt resistor in the negative lead to monitor the load current).

I uploaded a pictorial wiring diagram at http://i.imgur.com/UX4Y29o.jpg?1 for you.

In total about £600 (this was 2 years ago); £250 for the panel, £120 for the battery, £150 for the inverter (Ebay, 2nd hand) and the rest on sundries like cable & connectors (also Ebay in most cases). I used http://www.midsummerenergy.co.uk/ for a lot of the bits.

Note you can now get a 135W panel for £190, which is some indication of the falling cost over the last 24 months - the 50W Kyocera panel is now £80!

45W sounds like overkill for an RPi - I live in Scotland (55.95N) and I have a 50W panel charging a 110Ah battery in my garden shed. The battery is used to power a 1kW Invertek 250V inverter (150A fuses are impressive BTW) which I use to run a lawnmower, power tools, soldering station and lights (not all at once!).

The lawnmower is the biggest drain (900W) and is really pushing the envelope - but in the summer I can cut 55 sq. metres of lawn once a week without any charge problems. In the winter the grass doesn't grow :) but I can run the other tools just fine. I was very surprised at just how well this system worked, but then I did splash out on a high end 50W Kyocera panel when I designed this. My only alternative would have been to bury an SWA cable across common ground and somehow run it to my consumer unit in a 1st floor flat, so I was prepared to spend £300 (as it was then) on the panel. They're a lot cheaper now.