* Posts by cowbutt

22 publicly visible posts • joined 24 Jul 2009

The New ROM Antics – building the ZX Spectrum 128

cowbutt

Re: Paper Tigers

Huh? I used a Mannesmann-Tally MT81 dot matrix printer with my Speccy (va a Romantic Robot Multiprint Centronics interface). The output was rendered by the printer using its built-in fonts, and so looked the same regardless of the computer the text came from. I used Softechnics' "The Writer" word processor, which rendered to the screen at 40 characters per line, and scrolled horizontally so you could see the entire line. The printer was £99, and the Multiprint about £40.

No, working in IT does not mean you can fix anything with a soldering iron

cowbutt

Re: Customer pushback

I often repaired cables, headphones etc. when I was a broke teenager, but the repair was rarely as robust as before. So now I have the money to do so, I usually just "buy a new one".

SAP: It takes exploit devs about 72 hours to turn one of our security patches into a weapon against customers

cowbutt
WTF?

Re: Rock, meet Hard Place

I know testing feels like a good idea, but honestly, how much worthwhile testing can your organization do over and above that done already by the manufacturer? And, if you don't trust them, why the hell are you using their products?

Downloaded CCleaner lately? Oo, awks... it was stuffed with malware

cowbutt
Thumb Up

Re: Obligatory old geezer aside

And this is why, when I'm checking whether untrusted installers are signed, I check to see if they've been signed with the same certificate as previous, trusted installers.

It would be nice if the OS could pin certificates automatically, and highlight to users/admins if the signing certificate has changed from the previously-installed version.

cowbutt
FAIL

Re: "The dodgy software was signed..."

Symantec issue (sell) code-signing certificates to software authors. Those authors then sign their artifacts (executables, DLLs, whatever) before release, so as to allow users/admins (and the OS) to verify that they are indeed legitimate and unmodified. Symantec do not sign artifacts themselves, nor do they perform any kind of per-release code review of their customers' pre-release software.

Symantec (and other CAs) may do some due diligence to assure themselves that the applicant is not impersonating an established code signing entity, and that they are technically competent to keep their certificate safe and only sign things they intend to, but then again, they may not. Certainly, attackers have previously managed to get their hands on legitimate signing keys and certificates and use them to sign malware in the past - various versions of Stuxnet were signed with Realtek, JMicron and Foxconn certificates.

cowbutt
Alert

Re: Hopefully el-reg readers are aware of...

The 32 bit ccleaner.exe in the portable version of 5.33 included the Floxif malware. If you've ever run it on a system, that system should be considered compromised. You'll have to decide between Talos' recommendation to restore to a pre-CCleaner 5.33 backup of the system, or Piriform's assertion that Floxif only ever profiled machines and sent the details back to the attackers' C2 host.

The 64 bit ccleaner64.exe in the portable version of 5.33 seems to be safe. If you only ever used that version, then supposedly you have no reason to be concerned.

cowbutt
WTF?

Unless you have the resources and time to do analysis in a sandbox of every update that comes your way, automatic updating is still less risky than continuing to run software with known vulnerabilities. And, even if you do sandbox analysis, then there's still a chance that vulnerabilities in your existing version will be exploited before you complete the analysis to inform you that the update was indeed safe.

But, there's a logical problem - like looking for WMDs in Iraq, one cannot *prove* the absence of malicious behaviour: one more hour, day, or week of analysis might always turn up something unpleasant.

Smartphones merge into homogeneous mass as 'flagship fatigue' bites

cowbutt

I was burnt when, buying the then-flagship Samsung Galaxy S 2, I was shocked and disappointed by slow firmware updates, firmware updates with serious bugs that went uncorrected for months and sub-standard hardware (notably flash memory, but others also reporting problems with the BAT500 battery leaking and destroying the WiFi chip).

So my next phone was a cheap-but-solid SIM-free Moto G paired with a SIM-only contract. I figured if it lasted me a year before I needed to buy something equivalent, the two together are still only 3/4 the price of a flagship device at most.

Google gets my data, I get search and email and that. Help help, I'm being REPRESSED!

cowbutt

Re: Not so fast

Fair enough, but what's the alternative; assume that no-one has and have an - at best, paternalistic - government deciding what's right for everyone regardless of their competency in assessing the costs of any given transaction? No thanks!

US and UK declare red-team cyber war – on each other

cowbutt

My expectation is that the UK side gets some very-clever-but-not-terribly-strategic 0wnage of the US, but the US side comprehensively 0wns the UK.

Cameras for hacks: Idiot-proof suggestions invited

cowbutt

The Canon S9x/1xx compacts seem nice enough, but if you need a lot more zoom and better robustness, albeit with the downside of it being the same size and weight of a consumer DSLR plus kit lens, the Fuji X-S1 might also be worth a look.

Asus CEO sounds netbook death knell

cowbutt

Re: Shame

I was going to get one but everybody only seels 1gb ram versions

Pick up an N2800-based netbook (e.g. HP Mini 210-412x) while they're still out there; they can take upto 4GB (well, 3GB after the reservation for PCI IO, unless you're using a 64 bit OS. 4GB from Crucial was £15.59.

Home Sec: Web snoop law will snare PAEDOS, TERRORISTS

cowbutt

Re: This is insane.

Google "CNNIC certificate authority"

Essentially, GCHQ sets up a CA (or surreptitiously obtains assistance from one or more established CAs) and gets its root certificate installed in (i.e. trusted by) $ALL_THE_POPULAR_CLIENTS (IE, Firefox, Outlook, Thunderbird, K9, Chrome). Then, when they want to see what you're doing on Facebook, they issue a bogus certificate for a proxy they control and poison your DNS or use NAT to ensure you go via their proxy, rather than a legitimate Facebook server. You'll get the normal SSL "yellow lock" in your browser, and everything will look fine, but they can see (and optionally modify) anything sent and received.

Ten... top Android games

cowbutt

ScummVM

And whilst you're on an emulation streak, ScummVM to run LucasArts (and a few others besides) point-and-click adventure games.

Disk drive prices swell 5% every DAY in floods aftermath

cowbutt
Facepalm

"How the hell can you charge twice as much for something you already had in stick and got for a cheaper price that it would cost at the moment"

Because if it sells today, they'll want to replenish their stock so they can sell another tomorrow. Given the price rises, that replacement stock will cost rather more than what they bought the original for. If they can't sell another tomorrow, they'll not only lose the margin they could have made, but maybe also other trade as well if the customer goes elsewhere.

Travelodge still doesn't know who hacked it

cowbutt
FAIL

bulk email resenders have been targeted before

In October of last year I received spam to a number of semi-private mail aliases each used in connection with only a single web site. Eventually, I determined that each of these sites had used ThinkSend (aka createsend.com aka thinksend.com) so send their legitimate opt-in marketing emails at various times during 2009. One of the organisations followed up on this and confirmed that ThinkSend had been compromised during that timeframe: http://www.campaignmonitor.com/blog/post/2852/

More recently, I have received spam targeted at an address only known by me and laterooms.com, but their investigations drew a blank on that one. Thinking about it, I wonder if any data sharing goes on between laterooms and Travelodge?!?

Unused 'free' minutes cost Brits £800m a year

cowbutt
Boffin

Break-even

At 40p per minute for out-of-bundle voice calls, it's only 12 minutes before paying for them exceeds £5 for the next bundle up.

That was the process I went through (rather like Havin_it regrets not doing) before settling on a 300 minute/month tariff rather than a 100 minute/month tariff. I've never used more than about 120 minutes in a month - comfortably within 300 minutes, but would have cost more than £5 if I'd gone for a 100 minute bundle.

SanDisk flips out 32GB mobile phone card

cowbutt
FAIL

SDHC limited to 32GB

The SDHC spec is limited to 32GB, so support for >32GB cards will probably be hit-and-miss. SDXC goes to 2TB, but that requires different hardware, rather than just a firmware upgrade.

Crypto snafu grounds 3D Avatar screenings in Germany

cowbutt
Boffin

Film is old hat these days

Dolby 3D Digital Cinema (which is what the new digital cinemas seem to use, judging by the new Showcase De Lux in my city) is a digital format. I seem to recall reading that the cinema's servers retrieve the films via FTP and key material is delivered on DVD.

http://www.dolby.com/professional/solutions/cinema/digital-cinema.html makes for interesting reading.

Taliban attack Brit troops with explosive donkey

cowbutt
Grenade

Been watching Chris Morris' "Brass Eye"?

http://www.youtube.com/watch?v=_nvfQw8UCDE

Microsoft GPL violation hits memory hole

cowbutt

Re: Their own code?

Some Linux kernel symbols (functions, constants) are deemed protected and can only be used by kernel code that is GPLed (see http://lwn.net/Articles/205644/ , http://kerneltrap.org/node/4674 ). Code which uses these symbols is considered a derivative work and must also be distributed under the terms of the GPL. Code which doesn't use said symbols, isn't, and can be distributed under any licence. Presumably Microsoft's implementation of these drivers used some GPL-only symbols, and was not ported from some other OS either.

cowbutt
Linux

GPL Corrections

"Under the terms of the GPL, any module that's been combined with code licensed under GPL must be released under the GPL."

Incorrect; Microsoft could have chosen to cease distributing their code, and optionally re-written it in a non-infringing way.

Furthermore, the FSF has no right to sue in this case as the Linux kernel copyright isn't owned by the FSF. Only copyright owners whose rights have been infringed have the option of suing.