* Posts by Tom Wood

523 posts • joined 14 May 2008


You were told to clean up our systems, not delete 8,000 crucial files

Tom Wood

Re: xfer

I have a cron job that runs once a day and deletes files from ~/tmp (not /tmp) after 7 days and ~/Downloads after 30 days.

Stuff of a 'here's some notes I might need this afternoon' nature gets saved to ~/tmp. If it's something genuinely useful then it must be filed properly or hit the bit bucket next week.

London's Gatwick airport suspends all flights after 'multiple' reports of drones

Tom Wood

The internet

Can't these things be controlled remotely from the internet? How are they planning to find the operator when s/he could literally be anywhere?

OnePlus 6T: Tasteful, powerful – and much cheaper than a flagship

Tom Wood

Re: L500

I use my phone more than my personal laptop. It is more powerful than said laptop and crams that power into a smaller package... No wonder its more expensive.

Tom Wood

Re: My 5 is still good

My 3T is still going strong, as is my wife's 3 (though hers is on its fourth screen courtesy of the Nationwide's insurance policy).

Put your tin-foil hats on! Wi-Fi can be used to guesstimate number of people hidden in a room

Tom Wood


"Wi-Fi base stations could be used in shops, offices, hotels, cafes, and so on, to work out how many people are present at a given moment of the day"

Don't most of those places already have CCTV which could be used for such purposes... Probably more reliably?

Redis does a Python, crushes 'offensive' master, slave code terms

Tom Wood

Re: "Hurtful"

Like the old argument about 'brainstorming'. Someone decided that would be offensive to people with epilepsy and decreed it henceforth be known as a 'thought shower'. But then someone actually asked the people who actually were epileptic and they laughed: https://www.epilepsy.org.uk/press/facts/brainstorming-offensive

Google keeps tracking you even when you specifically tell it not to: Maps, Search won't take no for an answer

Tom Wood

Re: Firefox and DuckDuckGo

"Only sometimes?"

Actually it seems like a good idea to use more than one browser. And more than one credit card, more than one supermarket, etc.

Then at least none of them have *all* your data.

Tom Wood

Firefox and DuckDuckGo

A couple of weeks ago I found a lot of websites I use regularly kept crashing in Chrome on Android (repeated 'Aw Snap' messages). So I downloaded Firefox and found it works just as well as chrome (but without the errors).

While I was there I noticed it was easy to search DuckDuckGo instead of Google and indeed to set it as default. First time I'd used DDG and its actually pretty usable.

So, while Google still know my location etc they don't know what I'm searching for.

Also I noticed there were a lot of websites where audio recording and camera permissions popped up in Firefox - no idea why, they were just news sites etc - ads triggering these permissions? Needless to say I denied them as I have no need to use audio or camera from within Firefox.

So yeah, use a different browser sometimes.

Nah, it won't install: The return of the ad-blocker-blocker

Tom Wood

Re: Independent

Our local rag's website shows the first paragraph of an article then asks you to answer a survey question to see the rest of it.

Of course you don't have to actually read the question or any of the multiple choice answers, jabbing one at random works fine.

No idea who is paying them for the survey results but it can't be very useful data.

We shall call him Mini-U – Ubuntu reveals tiny cloudy server

Tom Wood

Re: also for Docker?

The article mentions containers in the first sentence and Docker in the third. So yeah.

Tom Wood

You can't have your cake and eat it

The things that make it comfortable to use also make it 'not small'.

Like how camping beds are less comfortable than regular beds.

Google Chrome update to label HTTP-only sites insecure within WEEKS

Tom Wood

Re: Shared Hosting

"no one can see what you're browsing with https" was never really true. The domain you are visiting is always sent in the clear during DNS resolution and the fact that you are connecting to a particular IP address is always visible (or the Internet protocol wouldn't work). Without SNI there was a one to one mapping from secured domain to IP address anyway.

What particular URLs you are accessing (below the domain level e.g. Pages within a site) is encrypted, and is still encrypted with SNI.

It's hip to be Square: Twitter founder Jack Dorsey's other firm targets White Van Man

Tom Wood

Re: I trust nothing Jack Dorsey touches...

But Square isn't really innovative, St least in Europe. PayPal, iZettle, Sumup and probably others all have type chip and pin and contactless readers and offer a similar service to Square. The square reader doesn't even seem to have a screen or a pin entry keypad.

Cambridge Analytica CEO suspended – and that's not even the worst news for them today

Tom Wood

Re: What I don't understand

Obama's (and yes, Corbyn's) campaigns mainly used social media in an open, person-to-person way, sharing links and personal opinions amongst their friends etc with the source clearly attributed.

What the Brexit and Trump campaigns did via Cambridge Analytica was paid, targeted advertising/propaganda which was not attributed to the party/group paying for it, and furthermore was targeted in an opaque way based on stolen data (they used the data of people's friends against Facebook's published terms of use and without their knowledge).

There is clear blue water between the two ways of using social media to campaign.

You can't find tech staff – wah, wah, wah. Start with your ridiculous job spec

Tom Wood

Re: Not knowing how to look can make it hard to find

Our interviews are very real-world, and of course still many people don't make the grade.

I'm sure I wouldn't. I have software deployed in C, C++, C#, Go, Python, SQL92, JavaScript, TypeScript and Kotlin on 4 different embedded ARM architectures plus Linux, Windows and FreeRTOS.

Can I remember the exact syntax, APIs, class libraries etc for all of them off the top of my head? Of course not.

Absolutely, and what about me saying our interviews are very real-world made you think I'd expect you to know all the syntax etc?

In the real world, people use Google. They look at books. They ask people for help. No, we don't let people use Google in our interviews but we do give people help and drop clues. It tells you more if someone can pick up on a clue and run with it rather than stare at you blankly (or argue that your clue is wrong). If someone says "I'm not sure but I'd look it up" I'd ask them what they would look up - it's no good searching Google if you don't know what you are searching for. It's no good reading an answer on StackOverflow if you can't understand it and tell the good answers from the bad ones - so we give people code to read and ask them to find what's wrong with it, and why. Would this code work? Could it be done better a different way? What do you mean by "better"?


Still, I've had more than one candidate who couldn't write a 'for' loop without help...that much I do expect you to know, in at least one language...

Tom Wood

Re: Not knowing how to look can make it hard to find

That's certainly not true of everywhere. A lot of what gets written about recruiting software engineers implies the employer is awash with candidates and it's a case of whittling the list down to the number of vacancies.

In my experience that's often not the case, at least in the regions we work in. Our interviews are very real-world, and of course still many people don't make the grade.

Tom Wood

Re: Bucolic programming

Interesting you should mention Bradford. We're a tech firm based on it's outskirts and we're certainly not alone in this region. And yeah, we struggle to recruit good engineers too.

Good luck building a VR PC: Ethereum miners are buying all the GPUs

Tom Wood

"Proof of work"

Greenpeace (etc) should award a prize to whoever can come up with a cryptocurrency where the "proof of work" is proof of /useful/ work - not just burning CPU cycles (= heating the planet/wasting electricity).

UK parliamentary email compromised after 'sustained and determined cyber attack'

Tom Wood

Re: Email != Webmail

Because the algorithms which are used to limit logins usually take into account the IP from which the attempt is made

Well don't use those algorithms then.

(I know, you could lock a legitimate user out of their account in that case, but maybe you could design some way to mitigate the impact of that, e.g. require a user to log in from a separate web system using decent 2FA or whatever to unlock their account in that case).

Tom Wood

Re: Email != Webmail

Why are limited login attempts not going to stop that happening?

Every time I change my work network password, I have to first stop my phone and email client auto-syncing with the server, otherwise I get locked out of my email for too many bad password attempts.

And if logins are automated, all the more reason for using long and complex passwords.

Tom Wood

Re: Why all the speculation?

So? Messagelabs provide the first MTA, I expect mail passes through a bunch of other servers before it reaches the users.

Sysadmin finds insecure printer, remotely prints 'Fix Me!' notice

Tom Wood

University printers

In our Computer Science department at uni (2003-2007) they had some network printers which were connected such that every page printed cost a few pence in credit. Students were allocated a couple of quid of free credit each term and if you used that up you had to buy additional credit using a coin box.

They also had stuck fixings to the printer trays with epoxy and used a padlock and chain to stop people fiddling with the internals, or taking paper from the paper trays.

Some bright spark figured out that you could walk up to the printer and print a diagnostics page from the menu, and it showed the printer's IP address. The printer had a FTP interface enabled and any postscript file transferred to it would be printed instantly. Of course this was all anonymous and bypassed the print credit charging system.

After that I don't think anyone seemed to run out of printer credits. One guy even took to printing a blank postscript file whenever he wanted a blank piece of paper.

Worse still, at least at that time universities had huge IPv4 address blocks meaning every PC - and yes, the printers - in the department had real-world public IP addresses. Not sure if anyone tried but I reckon you could have logged in to that printer from anywhere in the world, without any authentication, and printed stuff off.

74 countries hit by NSA-powered WannaCrypt ransomware backdoor: Emergency fixes emitted by Microsoft for WinXP+

Tom Wood

Re: When the dust settles...

The OS provider who has been warning for years 'this OS is obsolete, it's unsafe, we no longer support it, stop using it'?

Tom Wood

That looks like like the result of 'randomly' mashing the keyboard, not truly random. Lots of substrings formed of letters that are close together on the keyboard.

D'oh! Amber Rudd meant 'understand hashing', not 'hashtags'

Tom Wood

"Banning end-to-end encryption"

Haigh also asked what assessment has been conducted of the consequences for (a) the UK economy and (b) national security of banning end-to-end encryption?

How could the government ban end-to-end encryption? Or rather, how could they enforce such a ban? Assuming they can inspect all internet traffic, encrypted content should be indistinguishable from random noise, so are they going to ban sending random data over the internet? Assuming not, an encryption tool could just hide genuine data in amongst a stream of random noise; there's no way they could force you to decrypt the random noise because it's impossible to do so, and no way for them to tell which of the packets contain encrypted content.

Of course they could prevent named service providers from offering messaging apps that use end-to-end encryption, but someone (probably abroad) would just create another one, or the terrorists will use PGP or something. We all know how well attempts at banning that went.

Smart meter firm EDMI asked UK for £7m to change a single component

Tom Wood

Re: That doesn't sound ridiculous

A single developer working on a contract won't get that much, though they'll probably make a bit over half of that, with the rest being overhead costs (premises, recruitment, management, etc...) and profit for the company you are working for. An employee will make less but gets job security, holiday pay, sick pay, maternity/paternity pay, employer's NI etc etc.


Tom Wood

Re: I've had several phone messages left (home and mobile) asking if I want one

"But if I want something to reach a certain temperature, I want the rate of energy in to be as high as possible as all the time it's heating up, energy is also being lost to the surrounding area. So, in extremis, 100W for 10 hours won't get me to the same temperature as 10KW for 6 minutes, even though the total energy input is the same."

That's true if you're heating a fluid (e.g. boiling water for a cuppa) but not for a solid. If you cooked meat that way you'd end up burned on the outside and raw in the middle.

And gentler heating can be more efficient, for example you could cook a joint of beef in a big electric oven or a small slow cooker. The slow cooker would use less energy because it loses less heat to the surroundings, the cooker itself has lower thermal mass, it heats the beef directly by conduction rather than indirectly by convection, etc.

Tom Wood

Re: That doesn't sound ridiculous

I work in embedded software - not smart meters - but I've worked with on similar projects.

"Replacing one chip" is probably not as simple as it sounds.

From the hardware side - there will probably be board redesigns (operating in a different band may need new antennas, RF validation etc) which could result in several spins of hardware, refinements etc.

From the software side - we don't know what the interface is to this chip but quite possibly it will require entirely different drivers to the old chip. Maybe some kind of abstraction layer needs changing if the interface to the driver is different (we don't know if the new chip is from the same vendor as the old chip). Sure, drivers are often provided by the chipset vendor, but they also rarely work perfectly first time and need effort to integrate and test.

And then you have to test the thing, not only with the new bands but also with the old ones, and shake out all the weird edge cases, investigate, debug and fix them. And then there's the official certification process too, which is I imagine fairly onerous.

I'm not saying that this is definitely a £3.3m project, but in the absence of any other information, it certainly seems plausible that the engineering (and, yes, project management and QA) involved in such a project could come to that ballpark.

Tom Wood

That doesn't sound ridiculous

Works out an average rate of £98 an hour, which is not crazy. Depends a lot on where they are based and the availability of people with the right skills.

So the question is, are they over-estimating the resourcing needs?

Based on everything involved in designing and validating a hardware change and the software changes to go with it (does it need new drivers? How well tested is the new chip? Do you end up finding bugs and weeks of back-and-forth with the chipset vendor to resolve them? etc) 18 months and 13 people seems kinda the right ballpark.

Considering the outrage there would be if the project was rushed and the resulting code under-tested and buggy, maybe they are rightly estimating cautiously in this case?

New PayPal T&Cs prevents sellers trash-talking PayPal

Tom Wood

Re: If you only buy none of this matters

If you have a choice of paying by Paypal or by credit card directly, for purchases over £100, pay directly by card or you will lose the "Section 75" protection. http://www.moneysavingexpert.com/credit-cards/PayPal-Section75

Government hails superfast broadband deal for new homes

Tom Wood

Line rental

Line rental is not "unnecessary". You still want some kind of piece of string between you and "the internet" and providing that string costs money. Doesn't really matter whether you use that string to make analogue phone calls or not.

They should just call it a "standing charge" like the gas and electric companies do and have done with it.

New Euro-net will let you stream Snakes on a Plane on a *!#@ plane

Tom Wood

I'd much rather take a few hours off that I can spend at home the day before a long flight, knowing that I'll be able to get the work done while on the flight with nothing much better to do. There's not a lot you can do with your time on a long haul flight that is actually really relaxing, so might as well do work on the plane and relax at other times.

Tom Wood

I used American Airlines' inflight WiFi on a business flight from London to LA - I think they charged something like £15 and while it was by no means "superfast" the ability to spend a few hours of an otherwise long and boring flight working (sending emails, using VPN etc) was great and definitely worth paying for.

Only weird thing was that the service seemed to be tunnelled back to T-Mobile somewhere in Germany so you kept getting sent to German versions of websites.

UK.gov flings £400m at gold standard, ‘full-fibre' b*&%*%£$%. Yep. Broadband

Tom Wood

Re: 5, 6, 7, 8G...

There isn't (and can't be, due to limited frequency space) the capacity though. Certainly in an urban area, if all the data that flies about through cables was replaced with mobile you wouldn't be getting those good speeds...

EU €120m Wi-Fi spend explained, but not excused

Tom Wood

Ever read your own articles?

Read the first few paragraphs of this article and tell me what it's about.

"extending the reach of Wi-Fi" could mean anything, to me it sounds like some research to extend the range of a single access point.

But no, apparently it's to do with adding public Wi-Fi hotspots. Yet the word "hotspot" doesn't appear until paragraph 15.

Thanks, IoT vendors: your slack attitude will get regulators moving

Tom Wood

Time, cost, quality

As with most things, this is the classic trade-off.

Funnily enough industries such as aviation and nuclear spend a lot more money to find and fix bugs in their software than do people developing consumer grade software (desktop and mobile OSs, TVs, set top boxes etc). Consumers demand quickly-developed, latest and greatest software and it is neither possible nor necessary to deliver your mobile OS to the same standards of quality as you would the control software for a nuclear reactor. If you wanted your mobile phone to be as reliable as a warplane then (1) it would take decades to develop and (2) you wouldn't be able to afford it.

The same is true for other things, your house was not built to the same quality standards as the Channel Tunnel was because of the typical trade-off between time, cost and quality and the impact of failure. Software is no different.

And the IoT devices involved in this attack were bargain basement models made as quickly and cheaply as possible, therefore it comes as no surprise that the quality of their software is rock bottom (at least when it comes to security).

BT Yahoo! customers: Why! can't! we! grrr! delete! our! webmail! accounts!?

Tom Wood

EE, Plusnet: both owned by BT but for the time being at least operated independently, somewhat better

Virgin Media costs balloon by MEEELLIONS in wake of Brexit

Tom Wood

Re: They are a media company

Fibre cable, coax, headend equipment, ...

Colour us shocked: ISPs not that keen to sign up for Universal Service Obligation

Tom Wood

Upgrading to fibre is going to be one major way that customers will get a connection > 10 Mbps. Are you saying that if a customer has a 6 Mbps ASDL connection they can't be upgraded to a 50 Mbps FTTC connection until they and all of their neighbours has first been upgraded to a FTTC connection that has been throttled to exactly 10 Mbps?

Brexit Britain: HP Sauce vs BBC.co.uk – choices that defined voters

Tom Wood


is the version of the BBC website for those *outside* the UK. bbc.co.uk is for those inside. The former has adverts and the latter does not.

Try to access either from the wrong location and you will be redirected.

London's 'automatic' Tube trains suffered 750 computer failures last year

Tom Wood

Shows the challenge of creating self-driving cars

A railway is relatively self-contained, sure there are occasional junctions/points and the like but basically trains can go forwards or stop, and maybe occasionally reverse direction. It's rare to find people or other obstacles blocking the lines.

Not so a car on a road.

Milk IN the teapot: Innovation or abomination?

Tom Wood

Re: Never.

Actually, it scums up the coffee pot.

UK employers still reluctant to hire recent CompSci grads

Tom Wood

Re: I see nothing has changed

"Well I would say that someonee embarking on a CS degree should ALREADY know how to code......"

Well, that wasn't me. Sure I was "good with computers" at school, and learned a bit of HTML etc, and my A-level further maths included modules of discrete mathematics (algorithms and so on), but I didn't actually write my first "hello world" until I started my CS degree course in 2003. There were no programming courses at school, and nobody to encourage me to program.

I graduated with a 1st class MEng in 2007, and have done alright in software jobs since ;-)

I would imagine that most 17 year olds who think they know how to code, don't really. They may be able to hack together code from examples, but they probably don't understand the detail of *why* things are done the way they are (or better ways of doing things)

Tom Wood

Re: Interesting...

So maybe there really is a difference between "good" CS degrees and bad ones.

"A good way to test this is to give some example code that returns a pointer to a local variable and ask them to describe what can happen if you start to use that pointer."

We have a question based on exactly that problem in our interviews. Not only do I ask what happens if you use the pointer (it depends) but how they would locate such a bug in code that someone else had written.

It's not universally true, but of those with a few years experience, it seems that those without a CS background struggle more with questions such as this than those who studied CS, but that's just the impression I've got from the candidates I've interviewed.

Tom Wood

Re: unable to write a "for" loop in C/C++/Java

I can't remember the degree, but it was most likely computer science or something very similar (actually we've found at least a couple of unis have a course called Computer Games Programming which is actually basically Computer Science but made to sound sexier to 17 year olds who are applying for a degree).

Though, I'd still expect a Computing/IT graduate to be able to write a loop, maybe in Bash/Python/whatever scripting language they prefer but surely they're going to need to automate doing repetitive stuff at some point?!

Tom Wood

Re: unable to write a "for" loop in C/C++/Java

The guy who was completely unable to write a for loop was an extreme example, but the sort that sticks in your head. Sure, he'd probably done it before and I'm sure his uni software project must have contained many loops, but sat in front of a computer he couldn't remember the syntax. I can't remember whether he claimed to know C, C++ or Java but as the loop syntax is the same and we'd given him a choice of languages, the fact he couldn't do even this basic thing from memory was rather concerning. (It wasn't the only thing where he failed to show knowledge or understanding, however).

Tom Wood

Re: Interesting...

As someone who does interview both graduates and more experienced developers (and a comp sci grad myself), in an embedded software business, I'd say I most value someone who has learned the fundamentals of CS (algorithms, complexity, computer architecture, logic) and some software engineering (design, testing, OO, design patterns) and can evidence applying both through their project work. The "Android/IOS/Linux/Oracle/Windows 10/Azure/AWS" stuff I really don't care about, provided their project work shows they have applied some knowledge in some domain areas and can pick it up quickly. (Though understanding the basics of the Linux/POSIX style command line is a big plus).

One reason why I'm always wary of "experienced" programmers who were self-taught and came from a hardware or physics background for instance is that they can bash out code based on tutorials they've learned etc, but they don't really understand basics like what a pointer is or what the difference is between a list and a vector, for example. Which can lead to writing buggy software, or being unable to debug such issues in other people's code...

Tom Wood

Quality of graduates

We're a small/medium sized software consultancy (~60 or so employees in the UK). This year we advertised a vacancy for a software graduate. Many who applied who were either in the final year of their course or who had graduated with a Computer Science or similar degree failed our 10-question online multiple choice filter test. The test in question is open book and not time limited; the questions cover the basics of programming and CS theory, nothing complex; and our "pass" mark is only 6 out of 10. (Question 11 is "how many of the above answers did you look up online or ask for help with" - we wouldn't necessarily reject someone who looked up most of the answers, provided they got them all right!)

Of those who got to an interview (6 candidates if I remember correctly), none was up to standards (and our standards are not overly high for a graduate; we're talking basic failings like being unable to write a "for" loop in C/C++/Java). We left the graduate role unfilled this year. We do also take a "year in industry" student, who we interview about half way through their second year at uni, with the same questions and interview process, and universally the "year in industry" applicants were brighter and more capable than the graduate ones.

Which suggests that somehow we failed to attract the "good" graduates, and were left with a bunch who had somehow graduated or were on track to graduate in Computer Science but yet failed to understand the fundamentals of their chosen subject.

GOP delegates suckered into connecting to insecure Wi-Fi hotspots

Tom Wood

Re: Why? - Let's have some critical journalism

And actually, I don't need to control the DNS server, that just makes it easier. Since I can see and intercept all your traffic to my AP, I can look out for any initial non-HTTPS request and spoof a response, for example.

This also works with secure access points, if there is a common password I can get hold of (e.g. WPA2-PSK). If there's a hotel or pub that has a known WiFi password they provide to customers (maybe they stick it up behind the front desk/bar), for example, I could easily set up an AP using the same SSID and password and chances are at least some of the time (e.g. if your device has a stronger signal from my AP than from the hotel's) you will end up connecting to my network.


Biting the hand that feeds IT © 1998–2019