Robot cars: why I’m both excited and worried

Maybe this is selection bias, but most people I know seem very underwhelmed by self-driving cars. I am whelmed! I hope it works out.

I should clarify, because I discovered when talking to my parents about this that the term “self-driving car” isn’t self-explanatory to everyone. To them, it meant something like “slightly better cruise control”, and seemed very unexciting. It may be clearer to say “robot car” or “robot taxi”. A self driving car is a car that does every driving task by itself. It decides on the route. It looks where it is going. It turns corners. It brakes. You could, at the end point of the development, lie on a bed asleep inside a car while it drives you where it’s going. It’s still not entirely clear (to me) that this is all feasible with likely technology, or that it will be legally acceptable, but it’s seeming more likely. After years of driving modified cars with added self-driving about, Google is making prototype cars for further development.

Incidentally, “robot car” is a lot easier to type than “self-driving cars”, so I’m going to adopt it for the rest of this entry.

Upsides of robot cars (why I’m excited)

Here’s what I envisage:

Less driving by humans (specifically, me). Most people who comment on cars love driving, which is why I suspect self-driving car discussions often spiral into “but driving is fun! no one will cede control over their favourite activity!” Well, I’m queuing up. It’s not that driving is never fun, but for me, but it’s frequently unfun, especially since it’s so often city driving. Most of the fun bits of driving could be replicated for me in dedicated arenas.

Time reclamation. I drove to Canberra on the Easter long weekend, got stuck in the worst traffic I’ve ever been in, and commented to my husband that my brain is somewhat over-powered to be spent deciding when to slightly and briefly depress an accelerator pedal. Less human time spent driving is more time spent reading, talking, composing, sleeping. Plus all kinds of hedonism.

Increased independence for ‘dependents’. Google’s publicity already talks about elderly people who have stopped driving. But I know other people who can’t drive. I live with two of them. They’re 4 months old, and 4 years old, respectively. The degree to which it is safe or healthy for children of those ages to be left to supervise themselves in robot cars is debatable, but by the time they’re say, 7? Sure, the car could drive them somewhere for fifteen minutes or half an hour. At 14 or 15? They can tell it where to go. (15 year olds can’t drive in NSW, but when I was 15, I was independently mobile on foot and on bike, and occasionally in planes, trains and taxis.)

Two other people I’ve personally known who can’t drive had a seizure disorder and limited vision respectively, and there are lots of medical and psychological reasons that limit or prohibit driving, many of which would be compatible with being sole passenger in a car.

Speaking of psychological reasons, while I believe I am a driver of roughly average ability now and I now don’t seem to find it more stressful than others do, I hated learning how to drive, was petrified, and was a frustrating and frightening student. If I had equal mobility without ever having gone through that, I would not have. (And I think neither would my husband, who didn’t learn to drive until I taught him.) I know quite a few people in this category, including some unable to use cars to this day.

More comfortable trips. The possibility of being driven around while lying down or in a comfortable chair or, for that matter, while standing or exercising or drinking with friends. (Although see ‘Humans as cargo’ below.)

Fewer cars. Cars that can come and pick you up should increase utilisation of cars, as in, there will be less cars total, and less empty cars at any given time. This diminishes the use of mined resources, and has a carbon impact (the manufacture of cars has a substantial carbon footprint).

Lighter cars. With a vastly improved safety profile (which I am taking as read, otherwise I think the whole project is null and void) the weight required for modern safety features in cars can be ditched.

Safer cars. This is an assumption, but really, they’re not going to launch at all if they aren’t significantly safer. So if it happens at all, deaths and injuries in road travel should fall to near zero if robot cars become ubiquitous.

Land reclamation. Fewer cars means being able to reclaim some of the very significant amount of private and public land use currently devoted to parking cars.

Potential significant fuel savings. Inefficient human driving presumably has some direct fuel cost. In addition, robot cars can spread their usage more evenly over different routes, further saving time and fuel. Diminished vehicle weight saves fuel.

Lifting of speed restrictions. Robot cars can’t overcome some physical difficulties here (non-linear increases in power consumption with speed, increased braking distances with speed) but they can overcome the human error that makes high speed driving dangerous. Trips should become somewhat faster with a high density of robot cars on the road.

Downsides of robot cars (or why I second-guess myself)

Here’s the tweet that inspired this entry:

Even ignoring the possibility of political decisions inhibiting my utopia above, I think there are significant potential downsides, some of which are likely, some of which are hard to model.

Job losses. At the driving end: taxi drivers, truck drivers, chauffeurs, public transport drivers. If less cars were manufactured, a huge number of jobs involved in manufacture, in supplying manufacturers, and in supplying and selling cars. It’s likely the cars will tend to have reduced wear and tear, so mechanics and their supply chains would be affected. If road wear-and-tear and fuel usage is reduced, people employed in road maintenance and in the massive fuel industry are affected. All raw materials and their supply chain are affected.

Which is all very well for me, I’m not in the transportation industry. But we as a society suck at dealing with the people who are made redundant by technological progress, and we’re not showing signs of getting better at it even as more and more work is threatened with redundancy.

Resources use and carbon output may rise. With the increased ability of people to use cars (by adding children and medically unfit drivers, and some unwilling drivers, to self-directed car users) comes the risk of additional aggregate kilometres travelled by car with negative consequences for carbon etc.

Shortened lifespan of existing cars. If there was a fairly fast transition at any point, human-operated cars then in existence become obsolete fairly rapidly, which is a resources and waste nightmare exceeding the flat-screen TV transition environmental disaster.

Longer commutes. While commuting might be less miserable per minute, it’s entirely possible that commensurate increases in hours spent commuting might occur (or be demanded by the workplace), which for most people is a negative because of less time spent with family and friends, and a rather insidious negative at that (people apparently underestimate how bad it is, because it’s regular, something I originally learned about in Big house, big commute?, together with the idea of the “triangle of happiness”).

Unpredictable and likely bad disruption of public (cheap) transport. I don’t personally much buy into the intrinsic virtues of contact with strangers while travelling. (The “public” in “public transport”.) I don’t mind if opportunities to talk to strangers — or ignore them studiously in order to compensate for being in each other’s personal space — are diminished. I do mind access issues though. I would assume that the movement of currently unfit or unwilling drivers into robot cars would diminish use of public transport and therefore its availability. At the same time, I assume that individual robot car use will not be universally affordable (if primarily offered by for-profit entities as is likely). There’s no guarantee that, with presumed middle-class flight to robot cars, any longer distance transport will be available matching current public transport prices and coverage.

Unpredictable and likely bad availability of robot cars for less privileged users. If the safety profile is good enough, we might be able to eliminate child restraints in their current form, but there’s no guarantee cars will be available to fit wheel users, large people (tall, fat or both, although tall is more likely because it’s a male-associated trait), and large family groups. There’s no guarantee that non-literate people, or visually impaired people, or people who don’t speak the local majority language, will be able to operate the user interfaces. (I’d bet against, in fact.)

Privacy and autonomy. These are under substantial threat. Consider existing cases. Car sharing schemes track your location on GPS. Speed cameras record all passing vehicles, not just speeding ones. (Such data has been used as evidence in criminal trials in NSW.) Many transport smart card agencies hand over individual usage data (where you got on and off public transport) upon request to police, without a warrant. Robot cars have potential for all these abuses, plus the ability for other people to take over control of the car you’re in.

Known unknowns

There’s also some intensely unpredictable things:

Unpredictable disruption of places where people congregate. I do consider the existence of spaces serving as, essentially, town squares or village greens, a positive, and just as private cars significantly changed their location and availability, robot cars would too. I don’t know how and to what effect.

Humans become cargo. This is already more or less true in aviation, and it hasn’t played out so comfortably, with comfort tending to diminish in return for cheaper seats. But flying is fuel intensive (and the distances sometimes simply ridiculous). So it’s difficult to work out what Sydney to Canberra or Sydney to Melbourne would look like and it depends partly on power costs. Would a vehicle drive me to Canberra on its own? Would it go and hitch itself to some kind of road-train? Would a pod containing me be loaded onto a truck? Would I board a robot bus but otherwise have more or less the current bus experience? What would the effects on fuel use be of various models? (See I Spent 28 Hours on a Bus. I Loved It.) What would rich people do? What would poor people do?

Cargo becomes atomised. Conversely, when is the preference for large cargo loads (B-double trucks, trains full of coal) partly an issue of needing a human operator? Would those loads split into smaller units to be delivered on an as needed basis? When and at what point in the delivery process?

Increased drug use. Maybe? One inhibiting factor on drug use (the need to drive home) will be lifted. I’m not anti-drug use as a matter of principle, so this is not necessarily a negative to me.

Killing OWOOT

I founded Oceania Women of Open Tech (OWOOT) — a group for women in open technology in Australia, New Zealand and the Pacific Islands — in 2011. And today, after some discussion on its mailing list over the last week, I closed it due to lack of activity. I’m sad it didn’t continue in the way that all endings are sad, but I’m not especially sad. Not everything works out and I think it’s healthier to wind up things than to have everyone wonder what’s happening with them and whether any change is coming in the future. Especially when you consider the cookie licking problem. My moribund project could be inhibiting someone’s new project, simply by existing!

Speaking of my projects, or rather, things I founded, the Haecksen miniconference at is now separate from OWOOT, and continues to exist for the time being. (Although I am told attendance was low at 2014, and the current Haecksen volunteers might also reconsider it in a year or two.) Because there’s no longer an OWOOT list on which to coordinate it, I asked Linux Australia to found a new list. To help organise Haecksen in future, please join the Haecksen organisers mailing list. To attend Haecksen, watch the public Haecksen website or the Linux Australia mailing lists for information.

More fun with Fedora

Still going with Fedora 20 and my new X1 Carbon. I’ve decided it’s dangerously light. What if I got really annoyed? I could throw it quite far!

Linux does add something special to the new laptop experience after all, dodgy support for hardware for the first months I have a new laptop. I foolishly believed I had finished the install just because it booted and I could use a web browser and send email, but I kept noticing new problems as time went on.

First, the laptop’s screen (her name is Irian by the way, because I name them for female wizards, which, well, spoilers, but I do like Tales of Earthsea and you should read it) is very high resolution. GNOME tries to detect this and (essentially) make all its screen elements extra big to compensate. The trouble is, I usually use an external monitor which doesn’t have 2014-grade DPI (it’s 24″ with a , and GNOME doesn’t detect that, so things were being displayed on the external monitor extra big as well. This can be reverted with:

$ gsettings set org.gnome.desktop.interface scaling-factor 1

But that now means that when I disconnect my external monitor, everything on my laptop screen is eensy teensy. So switching between my laptop display and the external monitor involves plugging or unplugging a cable and a command line interaction. Judging from the discussion on bug 1025391, the task of figuring out when to apply what scaling factor is no mean feat, but all the same, it’s annoying that it’s been handed to me.

Second, speaking of external monitors, this is what happens when I boot now: the machine starts. My laptop screen goes black. My external monitor displays a featureless grey, ie, the background colour of gdm but without any content whatsoever (ie, it doesn’t display a list of users or a login prompt… just featureless grey). I have worked out that I can hit Enter, type my password, hit Enter again and then it will log me in, which is an improvement over the previous sequence which was “swear, pull out monitor cable, force a reboot, log in without external monitor, reconnect external monitor”. I’ve run Fedora 20 on my previous laptop and this didn’t happen, so I presume again there’s some specific hardware support issue where Fedora+gdm can tell I have an external monitor but not to the point of actually displaying a login prompt on it.

Meanwhile, I mentioned before that it wouldn’t resume from suspend and that I needed to upgrade the BIOS to get that fixed. There are a few ways to upgrade the BIOS within Linux but they all seemed horrifying, so rebooting into Windows was, in theory, going to be the way. However, as foretold in the prophecy, I didn’t have dual boot working in this new UEFI+Windows 8 utopia. In fact I still don’t, because I get a “cannot load image” error trying to boot Windows 8 through grub, and that error seems to either mean (a) you have Windows 8 and Fedora installed on separate physical drives (no) or (b) you shouldn’t be using grub but could maybe use one of a number of other bootloaders maybe because secure boot something something I don’t even. I messed around with this sort of thing until my eyes bled, and eventually resigned myself to going through the BIOS’s menus to boot Windows (ie, press Enter to interrupt startup, press F12 to get a startup menu, select Windows, it’s not that bad). I only use Windows for upgrading the BIOS and communicating with the Australian Tax Office on behalf of my business in any case (because you haven’t lived, died, and died again until you’ve tried to get the Australian government’s AUSkey authentication working under Linux, but I digress). I can deal with the BIOS menus for those cases.

The BIOS update went fine though, and now I can suspend and resume. And while I was in Windows random nagware immediately fired up to ask me to hand over my email address so as to confirm my subscription to a million anti-virus and anti-malware bundleware things that come pre-installed, so that certainly reminded me why it is that I don’t use it.

Standard disclaimer: I still don’t want tech support and no, I’m not going to file bugs. My Ubuntu-using days, in which you file a bug and the only response is having to extensively reconfirm it exists every three to six months for three years lest it be closed has pretty much cured me of bug-filing. I like shouting into the void on my own domain now.

New laptop blues

At a previous employer, my husband, who worked from home as a developer, was given a new laptop every three years, since it was his primary work tool. One of his colleagues, after going through the hassle of setting up a new laptop, apparently opined that he wished he was getting a new coffee machine or something similar.

Speaking of which, hello from my new Lenovo X1 Carbon, likewise my primary work tool! It’s amazeballs. It is the size I like (14″) while having the weight I’ve always coveted and previously associated with <12″ laptops (as weight little as possible of course, but 1.2kg or so is OK). I’m also joining the world of SSD drives, luckily modern spinning drives have way more space than what I use on a laptop (my photos are stored on an external drive, my music in FLAC on our central server and Vorbis or MP3 on our phones) so I didn’t even need to dial back in order to settle for 120GB. So far, so win.

But, oh, the setup.

First, I’ve never had a new laptop that entirely worked with Linux, and this one is no exception. It doesn’t resume from suspend (looks like this is bug 1084742 and I’m going to need to update the BIOS, so writing this entry has already paid off!). And sheesh moving my working environment from one laptop to another is a monumental pain. Especially when I’ve just reinstalled my Linode for the first time in about a decade, in order to install a 64 bit distro and thereby be able to use their SSD offering.

If you look for how to do such a thing on the ‘net, you get a few possibilities.

Use some kind of scripting/automation of the installer to get exactly the right packages, your config files set up the way you like them and such. I maintain a small number of Linux machines: three (hetrogenous) Ubuntu servers and a Fedora laptop. That’s, in my opinion, about three too few to find it worthwhile to, eg, semi-manually maintain a list of all the packages I need, work out the common versus custom bits of their config files, and such-like, especially when I reinstall so seldom. By the time a reinstall comes around, I can guarentee you I will have accidentally busted my automated install config through lack of testing, or the entire software stack I was relying on for the automation has been discontinued for years.

Copy /home and /etc to the new machine. Yeah, don’t do this.

Well, /home is basically OK, as long as you check the user ids carefully. (Fancy that, some people still run multi-user systems.) But don’t wholesale copy /etc. It worked OK for the Linode, once I edited /etc/fstab to mount the new drive configuration and chowned a bunch of things in /var to account for some of the user ids changing. Which was silly of me and which isn’t really what you’d call working, but it works now.

It was a monumental disaster on Fedora though, because I don’t speak new-fangled Linux. Specifically, I have no idea how one mounts LVM partitions from the command line and had to rely on Nautilus for it, and it turns out that if I, eg, move a new file in over the top of /etc/postfix/, SELinux won’t let it read it any more and I have to either understand SELinux or invoke random magic commands found on Google that probably amount to “disable SELinux and mail my SSL private key to the NSA while you’re at it”. Or I could understand LVM and SELinux of course, and that would be what I’d do if rebuilding a laptop wasn’t a 3–5 yearly task for me. Once again, whatever I learned will be thoroughly out of date by the time I next need to apply this knowledge.

And separately, there’s the package installing problem. Basically, both Debian-verse and Red Hat-verse systems both now have package managers that track the difference between “this package was installed at the administrator’s request” and “this package was installed as the dependency of another package”. But neither of them, as best I can tell, can export this reliably to a second machine, which means that on my new Fedora laptop, both Firefox and libwhatever.something.the.millionth are treated as sacrosanct “installed at the administrator’s request!” packages and I’m stuck with libwhatever.something.the.millionth forever, because I used rpm -qa. (There’s attempts at getting only the right packages out of the package manager, and the leading solution is now busted to the point of giving about 200 errors and then telling me I’d only ever installed 10 applications on my old Fedora install. You see what I mean about this stuff aging.)

Use some other operating system. Judging from commentary on the “yay, a new lapt— shit, a new laptop, now to spend three days of my life spinning my wheels on reinstalling all my favourite apps and redoing all my config” situation I’ve heard from Windows and Mac-using peeps, I get the impression this is a universal problem.

Use some magical program where one points at an existing laptop and say “make it like that one!” Dreamland. Although you’d think it would be something of a market advantage for Linux, which typically is agnostic on which packages you use (as long as they are open source and have certain trademark properties, admittedly, browsers are an issue here).

But I’d use the hell out of a desktop replicator, if one existed. Or even something that reliably dumped my package status including the “installed as a dependency” distinction, plus gave me some hints as to which bits of /etc I probably want.

Standard disclaimer! I’m not after any of: requests for further information for debugging purposes, exhortions to file bugs, or explanations of how to do anything with LVM and SELinux. I can figure out where to look that up when hell freezes over or it becomes a paid job of mine, one or the other.

Federal election minus 1 day: last minute Coalition announcements special!

This article originally appeared on Hoyden About Town.

Mostly via tweeps, info on certain last minute Coalition announcements.

First, people saw the original version of The Coalition’s Policy to Enhance Online Safety for Children (original version), which read:

We will work with mobile phone companies (such as Telstra, Optus, Vodafone and their resellers) to develop online safety standards for smartphones and other devices with mobile network connectivity such as tablets, applicable to their use by children in two age groups: children up to the age of 12 years and teenagers.

As has recently been achieved in the UK, we expect these standards will involve mobilephone operators installing adult content filters on phones which will be switched on as the default unless the customer proves he or she is at least 18 years of age.

The Coalition will work with internet service providers (which provide fixed line broadband services to the home) to develop online safety standards for those services, recognising that they are very often accessed by children.

As has recently been achieved in the UK, we expect these standards will involve the major internet service providers providing home network filters for all new home broadband services, which will be switched on as the default unless the customer specifies otherwise.

This is a very different approach to the discredited compulsory filter proposal championedby the Rudd-Gillard Government, which was abandoned as unworkable.

The Coalition’s approach aims to empower parents — by giving them the choice of whetheror not to operate a filter at home, but by establishing the default setting as one which provides maximum protection.

The Coalition’s Policy to Enhance Online Safety for Children original/repudiated version, page 7

The Coalition very quickly backed away from this proposal:

Mr Turnbull quickly released a statement to clarify the Coalition’s position.

“The Coalition has never supported mandatory internet filtering. Indeed, we have a long record of opposing it,” the statement said.

Malcolm Turnbull’s statement is available in full on the Liberals’ site. The revised and/or correct version of the policy (depending on if you believe that they did accidentally make an early version public), reads:

Wewill work with mobile phone companies (such as Telstra, Optus, Vodafone and their resellers) and internet service providers (which provide fixed line broadband services to the home) to make available software which parents can choose to install on their own devices to protect their children from inappropriate material.

This is a very different approach to the discredited compulsory filter proposal championed by the Rudd-Gillard Government, which was abandoned as unworkable.

The Coalition’s Policy to Enhance Online Safety for Children current version (as linked from the Liberals’ policy listing), page 7

Moving on to last minute announcements they haven’t backed away from:

A Coalition Government, if elected, will crack down on Labor’s addiction to waste by auditing increasingly ridiculous research grants and reprioritising funding through the Australian Research Council (ARC) to deliver funds to where they’re really needed.

Some of the grants issued by the ARC in recent years have been, frankly, completely over the top.

There will be no reduction in research funding. In fact, the Coalition has announced new research into dementia and diabetes.

The Coalition would look to targeting those ridiculous research grants that leave taxpayers scratching their heads wondering just what the Government was thinking.

Taxpayer dollars have been wasted on projects that do little, if anything, to advance Australians research needs. For example:

  • The quest for the ‘I’ – a$595,000 grant aimed at “reaching a better understanding of the self”;
  • $160,000 on an examination of “sexuality in Islamic interpretations of reproductive health technologies in Egypt”;
  • a $443,000 study into “The God of Hegel’s Post-Kantian idealism”; and
  • $164,000 for a study into “how urban media art can best respond to global climate change” .

Ending More of Labor’s Waste, Liberal press release, September 5 2013

(My Honours supervisor is out there jumping for joy at the punctation used for that list, at least.)

Several research and research-affiliated groups have denounced the policy:

  • Jeannie Rea, National Office, The National Tertiary Education Union: This is a direct attack on the academic freedom of researchers working in Australian universities. If Tony Abbott wins, independent research loses.
  • Catriona Jackson, CEO, Science and Technology Australia:

    Specific research projects – all in the arts and social sciences – have been labeled increasingly ridiculous. But scientists know that the flow of new knowledge is critical to the kinds of real word results that all Australians are proud of, and that the Coalition is calling for.

    It was CSIRO scientist John O’Sullivan’s search for exploding black holes that led to his discovery of wireless technology that has swept the world, and earned Australia $500 million in royalties with probably as much again to come.

You can also review STA’s overview of science policy for the Federal Election, if interested. The Conversation has a couple of pieces about the policy, one noting that we already have a body of expert scrutineers who reject any number of bad and merely only very good grant proposals, and we call it the Australian Research Council, and the other analysing the craftiness of the press release itself.

What else have you seen sneaking in under the radar now that the advertising blackout is in place?

Front page image credit: Election Day CC BY-SA David Morgan-Mar, from the 2007 Federal election.

Now brought to you by Fedora*

I’ve been an Ubuntu user since about September or October 2004. I bought my first up-to-date laptop hardware in New York City (a Fujitsu Lifebook, still my favourite of my laptops), replacing a Toshiba Libretto I’d bought in late 2002 or early 2003 at more than five years of age and which I’d managed to squeeze Debian onto against its will. In 2004 my husband was working for the company later to be known as Canonical and so I became a beta tester (I think not a highly contributing one) for the distribution soon after revealed to be Ubuntu. And that was pretty great for me, basically Debian with a regular release schedule centered around up-to-date GNOME.

In January this year I appeared on My Linux Rig and you can see I was still an Ubuntu desktop user. I wrote:

I am curious about how Fedora is doing these days, but realistically switching distributions is more work than upgrading Ubuntu so I am likely to stick with the path of least resistance.

But rumblings were changing my mind. Late last year I made a belated upgrade to Ubuntu 12.04 (after I submitted my PhD in May), at which point for reasons I now forget it became impossible to use GNOME 2/Metacity. I wasn’t particularly enamoured of GNOME 2 by that point in any event, but I’d resisted switching because my husband has been using Unity for considerably longer (he is a fan; he may have been dogfooding for Canonical fairly early, although he’s worked for Google since mid-2011 and I am not sure of Unity’s timeline there) and I really struggled with it when I used his machine. Much later it emerged that he doesn’t use workspaces at all in Unity, so that may be responsible for his desktop being a bit Mary-hostile.

I gave Unity and GNOME Shell about two hours each on my desktop and decided that I liked the latter better. GNOME Shell wasn’t ideally supported in Ubuntu 12.04 and 12.10 but it worked well enough to keep me from the pain of re-installing. But then I upgraded to 13.04, and GNOME Shell crashed about every half an hour on my hardware and graphics seemed unstable in general. Unity was rather better, needing a restart “only” a few times a week. But I really missed GNOME Shell. I was tempted to move to a distro that follows mainline GNOME at that point, but the decision was sealed when I began to learn about Canonical’s plans for the desktop stack. I don’t actually have a strongly held opinion on a lot of the issues: the value or otherwise of collaborating with upstream in general or with GNOME or Wayland or Xorg in particular, the relative technical merits of any current proposal, the risks of splitting the Linux desktop and so on. I just have a preference for vanilla GNOME 3 and Canonical’s development direction suggested Ubuntu was increasingly less likely to cater to me as time went on. And less likely looked pretty bad when 13.04 already rendered it nearly unusable.

Well, I guess I do have a preference in a way, I’m using Fedora — rather than any other distro with a good GNOME 3 stack — to support Red Hat (in a small way), in that they are active in developing the software I like at the moment.

In terms of work, I really didn’t want to switch. Reinstalling my machine and setting up my work environment has been exactly as annoying and boring as I expected it would be, I have a whole second post coming with notes on all the gotchas I encountered configuring Fedora. There is nothing fun about installing or configuring Linux, and FedUp better do what it says on the tin and take me to Fedora 20 and so on when the time comes. (Ubuntu’s preferred upgrade path, by the by, hadn’t worked for me for at least five releases, I was therefore still using apt-get dist-upgrade.) It took me a month to get from “I want to switch to Fedora” to actually installing it, and it probably would have been at least another month if Unity hadn’t crashed on me about three times in an hour last week.

So here we are. Initial signs are promising. My install, while boring, went cleanly. GNOME 3 on Fedora is much more stable than GNOME 3 or Unity on Ubuntu 13.04 on my hardware.

Hopefully I won’t be doing this again before 2022.

* Not really, my servers are still Ubuntu LTS and will likely stay Ubuntu LTS or, if there’s some Unity-equivalent disruption in the Ubuntu server experience, which I can’t imagine, Debian.

Filling a car: user interface commentary thereon

I don’t own a car, so while I’m a bit late in life for this tradition, I’ve nevertheless been driving my father’s car while my parents are overseas. They’re back today, so last night I decided to fill the tank for them before they got back.

I wasn’t coming into this in the best of states. I had a three year old child in the car. It was evening peak hour in Sydney, and although I was yet to realise that events in Moore Park were slowing traffic even more than usual back as far as the Lane Cove tunnel (for reference, Moore Park and the Lane Cove tunnel are 15km apart on entirely different sides of Sydney Harbour), I had already had to turn from Lane Cove Road onto Epping Road, which has to be one of the worst designed intersections of all time, except for all other intersections of major arterial roads in Sydney, which are also awful in peak hour. (For example, it was often considerably faster to get off my bus on the Pacific Highway, walk 1km around onto Epping Road, and catch an entirely new bus further ahead in the queue than it used to be to wait for the bus to turn that same corner.) But Lane Cove and Epping is my especial enemy after most of a decade at Macquarie University, I can’t even go into it now. And finally, I was late to meet my sister, who was sitting on the front step of my house in the dark.

Then I pull up to a pump, which is also (I knew) on the wrong side of the vehicle, run back and forth between the drivers seat and the fuel hatch (on opposite sides of the vehicle) until I find the latch for it, unhook the hose from the bowser, drape it over the top of the car, and get a good look at the fuel cap for the first time. “DIESEL”.

Before everyone reaches for smelling salts, all that happened here is I said “oh for real?”, put the ULP hose away, got back in the car, moved it, hunted around on foot for the diesel pump, found it, moved the car there, filled the car, spilled big splotches of diesel all over my dress (that made for a fun drive home, ugh, sorry your car interior smells of diesel Dad, but I also note it smelled strongly of cattle before that), paid for the fuel, got back in the car, apologised profusely to my 3 year old — who is very well behaved in cars, those of you who’ve heard my story about him in planes will be surprised to hear, and who hadn’t peeped the whole time other than to say “oh no Mama diesel” sympathetically — and drove home in infuriating traffic, about 45 minutes late to hand over the car to my sister.

So far so good right? But my point is this. That label “DIESEL” was in a nice elegant thin font in white letters on the fuel cap. It was big but it didn’t look so terribly important, I can imagine “TOYOTA”, say, being lettered much the same (or “NO SMOKING” which is important in general, but less so to me in particular). I probably only would have needed to have been in about a 10% worse mood to have just missed it entirely and filled the tank with ULP, which I just now confirmed is as expensive a mistake as I thought it was, and this morning my parents would be flying into the country in order to find that I’d wrecked the engine. Good grief.

My point is this: it would be nice if that cap was, say, all in red, and burned to the touch in the close proximity of ULP or something (yeah yeah, not really). In order to avoid a mistake that would cost weeks and ten thousand dollars to rectify, and moreover would be at the expense of my father’s very car reliant job too, there’s elegant white lettering on black? There aren’t even differently sized or shaped interfaces? At least I can take a UI design lesson from it: I will always in future imagine evening peak hour, a toddler, running late, and how to help that person not spend ten thousand dollars on a momentary oversight.

And if you have a diesel vehicle and want to loan it to your frazzled adult daughter (or frazzled adults of your acquaintance in general) I see that there are after market mis-fuelling prevention devices. Good to know someone stepped in. Although at this particular service station, I would have had to pull it off again because it was a high flow bowser. So, you know, not exactly ideal still.

Why my phone is silent during LCA talks

I don’t especially like Tasker’s interface, but setitng one’s phone to silent is nice enough to bust it out, so I thought I’d explain how I do this during

A bit of background: Tasker is an Android application (not free in either sense of the word) that does things to your phone when certain conditions (called contexts) are true. For example it could change the wallpaper (task) when you have unread text messages (context). I have, for example, Tasker tasks that turn my phone to silent between 10:30pm and 7:30am local time; and to run rsync backup (which copies the contents of my phone to my home server, ie backs it up) every time it is both on power and connected to my home wireless network.

Tasker somewhat trades between UI simplicity and power in favour of power (although even then I think there are better possible UIs for it). You can generally find specific apps that do individual Tasker-like things (for example, I would not be surprised if there was a ‘Silent at Night’ app), but Tasker lets you specify a wide variety of contexts and tasks.

First: the LCA calendar iCal is in my Google calendar, so it’s available to Tasker through its Calendar contexts. So that’s prior to setting this up.

The basic setup would be this:

  1. Go into Tasker.
  2. Add a Context (called eg ‘LCA activities’), select ‘State’, ‘App’, ‘Calendar Entry’.
  3. In Calendar Entry, go down to Calendar, press the search icon, select your LCA calendar.
  4. Press the tick.
  5. Now it will prompt you for the task, which is silencing your phone. Select ‘New Task’. Name the task (‘Silence’): it might be useful for other contexts!
  6. Press + to add an action. Select ‘Audio Settings’ and then ‘Silent Mode’. Turn ‘Mode’ to ‘On’. Leave ‘If’ alone. Press tick to approve the action and then tick to approve the task.

After this teeny (ahem) amount of work you now have a Tasker task that silences your phone during any event on the LCA calendar.

Fine print

My setup is a bit more complicated than this because I thought ‘wait, I want my phone to ring during meals’. This is a pain in the neck to do.

I added a second Context (long hold on the existing context), another Calendar Entry, also on the LCA calendar, but I also searched for location, selected ‘MCC Foyer’ (which is where the morning and afternoon teas are) and selected the Not tickbox, to make it a negative context. The total effect is that when there’s an event in the LCA calendar AND when there’s not an event in the LCA calendar that is in MCC Foyer, the task triggers. But that’s quite a bit nastier.

It can end up being easier to have a calendar that amounts to a ‘Do Not Disturb’ calendar, which isn’t ideal. Some people do something like “silence during anything in my work[/personal] calendar that’s marked busy”, etc etc, which would be longer lived than my LCA recipe. BUT at least my LCA recipe buys us silence for this conference!

Creative Commons License
Why my phone is silent during LCA talks by Mary Gardiner is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Teach me py.test (Haecksen miniconf, Tuesday)

I didn’t manage to go speaking-free for LCA2013 after all, because I have volunteered to help out my roommate Brianna Laugher with the py.test presentation in the Haecksen miniconf.

The plan is that we will do “Teach me py.test” along the lines of Steve Holden’s “Teach Me Twisted” session at PyCon 2008 (see Catherine Devlin’s report). The idea of the session is that I (genuinely new to py.test, although not to either Python or to unit testing in general) will hook my laptop up to a projector and learn how to write tests in py.test, with Brianna teaching me.

We have pulled some of the business logic out of Zookeepr into this git repository in preparation for the talk at 16:05 in MCC6. I am not sure how much we will cover in 25 minutes, presumably not a lot, but it should be an interesting experiment in presentation style.

Fun at LCA 2013: my picks for Thursday and Friday


I rarely go to LCA’s tutorials, but really, after years of not having to worry too much about distributed version control systems due to having in-house technical support from my husband, a (now former) Bazaar developer, it’s probably time that I came to grips with git. Hence Git For Ages 4 And Up (Michael Schwern) is tempting, hopefully it’s OK for those of us who do use terms like “directed acyclic graph”. This does mean missing Wiggle while you work (Neil Brown) though: apparently you can’t be a git beginner whilst being interested in newfangled patching algorithms.

After lunch The IPocalypse – 20 months later (Geoff Huston) calls to me: it’s the sequel to his LCA 2011 keynote, which is the one that stood out to me. (Well, and Mark Pesce’s, yes, but funnily enough his actual content largely passed me by.) All that doom and gloom, and now what? Has IPv6 cost us our Internet?

A Tridge talk (Building a free software telemetry radio system) is an even more obvious pick than a Matthew Wilcox talk. (Although why did we put that particular talk up against Buffer Bloat? Tridge is going to talk about TCP performance issues.)

In the afternoon Keith Packard has a new passion (Teaching Robotics and Embedded Computing with Legos and Arduino) and then Ristretto: run-time types for JavaScript (Shane Stephens) sounds alarming. In a good way.


It might also be a two-tute LCA, with Beyond Alt Text: What Every Project Should Know About Accessibility (Denise Paolucci) up first. BUT NovaProva, or How I Did Six Impossible Things Before LCA (Gregory Banks) is the good crack (“NovaProva implements true reflection in C/C++”???), so… difficult!

After lunch, Asheesh Laroia’s Quantitative community management is closer to what I do but I am also curious about The real story behind Wayland and X (Daniel Stone). In the final session, probably Building Persona: federated and privacy-sensitive identity for the Web depending on how my conference energy is going.

And then where?

I’m headed back to the USA in March for PyCon, and I’m looking forward to having way (waaaaaay) less commitments than I did at Wikimania 2012, and therefore being able to catch more of the talks. And not dragging myself to my hotel room at 4pm to order crème brûlée room service because I am too tired to figure out how to work the lifts. (It was good crème brûlée though!) The Ada Initiative will probably be running some non-talk activities though, so it won’t be wall-to-wall talks. And then a second return to the USA for AdaCamp SF. And that really might be enough for one year, but if not, there’s always Kiwicon.