Telnet to your server:

telnet 127.0.0.1 11211

List the items, to get the slab ids:

stats items
STAT items:3:number 1
STAT items:3:age 498
STAT items:22:number 1
STAT items:22:age 498
END

The first number after ‘items’ is the slab id. Request a cache dump for each slab id, with a limit for the max number of keys to dump:

stats cachedump 3 100
ITEM views.decorators.cache.cache_header..cc7d9 [6 b; 1256056128 s]
END
stats cachedump 22 100
ITEM views.decorators.cache.cache_page..8427e [7736 b; 1256056128 s]
END

Thanks to Boris Partensky in the Memcached group here

There you go!

via the BootUp Labs Blog.

• Graham King on identi.ca
• Graham King on Twitter

The nature of the medium means those thoughts will be generally raw and truncated, but timely.

Last chance to invest in a firm favourite

They are covered again in more detail in this paragraph:

The x y z Fund only launched six months ago, but has already attracted considerable interest. To keep it small and flexible the number of units has been capped at 200 million. Last week they had reached two-thirds of that total and interest is intensifying. In the last two days alone they sold over 6 million units, so it is likely to close very soon.

Social Proof

The first principle is Social Proof. This fund is “a firm favorite” and “has already attracted considerable interest” and already “sold over 6 million units”. In other words, other people think it’s a very good idea to invest in this fund, so you should to.
Many of the top researchers in the field of influence (social psychology, behavioral economics, etc) worked with the Obama campaign, Their last minute get-out-the-vote message was “A Record Turnout Is Expected”. People are more likely to vote if they think other people will vote. The most beautiful part of social proof is that it is undetected – most people will deny that this is the case.

No-one know which fund to invest in, because most of them under-perform the index, and if the stock market really is a random walk, then there is no sane way to invest. It is in exactly that type of ambiguous situation where people look to other people to decide how to react.

Scarcity

We want what we may not be able to have in the future, that’s the Scarcity Principle. This is the “Last chance” to invest in this fund because it is “likely to close very soon”. Fear of losing out on something can be an extremely powerful motivator, and you see it every day when stores have end-of-line or closing-down sales, or when something is offer in ‘limited quantity’.

Invest with your calculator, not your heart

That email is an emotional appeal, not a rational one. Selling a new fund during a recession must be difficult, and there’s no reason to think this one is selling well. The language doesn’t promise anything concrete, so it doesn’t lie. It just influences. And that is good copy writing.

The seminal book on this topic is Robert Cialdini’s “Influence”, which I highly recommend. It’s a New York Times best seller, and Fortune Magazine lists Influence in their “75 Smartest Business Books.” You see what I did there? :-)

Luckily for me, message queues are this years Hot New Thing, so there’s some good options. I looked at RabbitMQ, Gearman, Beanstalkd and StompServer.

I’d like the message queue to play nice with Python, with Ubuntu, and take almost no memory, as I’m on a Virtual Private Server, and I’d like it to stay up forever. I want small and solid.

Summary

Memory size is the resident set size, obtained like so: ps -Ao pid,rsz,args | grep <name>. If there is a better way of estimating memory please let me know in the comments.

RabbitMQ

An all-singing all-dancing “complete and highly reliable Enterprise Messaging system”. With language like that you’d expect horrible bloat and per-cpu licensing, but happily that’s not the case. It’s straightforward to setup and relatively lean.

The protocol, AMQP, comes from the financial world, and is intended to replace Tibco’s RendezVous, the backbone of most investment banks. There’s lots of documentation, lots of users, a healthy ecosystem, and it looks good on your CV.
I tried RabbitMQ first, and liked it so much I almost stopped my evaluation right there and deployed it.

The best tutorial for using it from Python is here: Rabbits and Warrens

Publisher

Quick Code

Consumer

Quick Code

Gearman

Gearman is a system to farm out work to other machines, dispatching function calls to machines that are better suited to do work, to do work in parallel, to load balance lots of function calls, or to call functions between languages.

Developed by Danga Interactive (essentially Brad Fitzpatrick, who brought us Memcached and Perlbal). Used by LiveJournal, Digg and Yahoo.

The original Perl version is the Ubuntu repository as ‘gearman-server’, but that’s not the one you want. To get the newer leaner C version follow the instructions here: https://launchpad.net/~gearman-developers/+archive/ppa You’ll also need to add the key:

• sudo apt-key adv –keyserver keyserver.ubuntu.com –recv-keys 1C73E014
• sudo apt-get update

Client

Worker

Beanstalkd

Beanstalkd is a fast, distributed, in-memory workqueue service. Its interface is generic, but was designed for use in reducing the latency of page views in high-volume web applications by running most time-consuming tasks asynchronously.

Developed for a very popular Facebook Application. The smallest memory footprint: after startup, connecting, sending a few messages, it’s resident memory size (rsz) was still only 0.5 Mb!

To install the server:

• sudo apt-get install libevent-dev
• wget http://xph.us/dist/beanstalkd/beanstalkd-1.3.tar.gz
• tar xvzf beanstalkd-1.3.tar.gz
• ./configure
• make (there’s no install step, it just generates the file ‘beanstalkd’)

To install the Python library:

• wget http://pybeanstalk.googlecode.com/files/pybeanstalk-0.11.1.tar.gz
• extract it
• sudo python setup.py install

There’s a good tutorial here: http://parand.com/say/index.php/2008/10/12/beanstalkd-python-basic-tut orial/

Producer

Quick Code

Consumer

Quick Code

StompServer

StompServer is a lightweight pure Ruby STOMP server.

To install the server on Ubuntu:

• sudo apt-get install ruby-dev rubygems
• sudo gem install stompserver

To install the Python library:

• wget http://stomppy.googlecode.com/files/stomp.py-2.0.1.tar.gz
• extract it
• sudo python setup.py install

There’s a good Python / Stompserver tutorial here: http://morethanseven.net/2008/09/14/using-python-and-stompserver-get-s tarted-message-q/

Sender

Quick Code

Listener

Quick Code

Results and Conclusions

I’d be happy working with any of these four. All four were easy to setup, fast, decent in memory consumption, and had good Python libraries.

RabbitMQ has the most mindshare (it is the only one which registers on Google Trends), but it took the most memory and is the most complex to use. It looks like a great product, but it’s Message Oriented Middleware, not an in-memory job queue, so it’s not what I’m looking for.

StompServer had the least documentation, and took several times more memory than Gearman and Beanstalkd. In seems the most immature project, but would probably be a good choice for someone working in Ruby.

Beanstalkd is great. I would like to see it in the Ubuntu repositories, and it’s Python lib in PyPI, but aside from that, I can’t fault it. I’m not choosing it, because Gearman is even better.

Gearman was designed for exactly the problem I have, takes almost no memory (1.4Mb), has a great pedigree (Danga), is widely deployed (LiveJournal, Digg, Yahoo), is in Ubuntu (almost), has a Python library is in PyPI, and someone helped me out on the #gearman IRC channel straight away. It even has queue persistence and clustering. So, Gearman it is.

In response to monkeys stealing his coffee beans, an Indian farmer observes: If you start shooting monkeys, you’ll spend the rest of your life shooting monkeys.

via Bruce Eckel

http://www.nytimes.com/2009/07/19/technology/19distracted.html

Here’s some excerpts:

in a survey of 1,506 people last year by Nationwide Mutual Insurance, 81 percent of cellphone owners acknowledged that they talk on phones while driving, and 98 percent considered themselves safe drivers. But 45 percent said they had been hit or nearly hit by a driver talking on a phone.

That’s the Lake Wobegon effect, the tendency for overestimate their capabilities in relation to others.

…research, showing that multitasking drivers are four times as likely to crash as people who are focused on driving, matches the findings of two studies, in Canada and in Australia, of drivers on actual roads.

The highway safety administration estimates that drivers using a hand-held device are at 1.3 times greater risk of a crash or near crash, and at three times the risk when dialing, compared with others who are simply driving. The agency based its conclusions on research from the Virginia Tech Transportation Institute, which placed cameras inside cars to monitor drivers for more than a year. The study found cellphones to be the most common cause of driver distraction.

Research also shows that drivers conversing with fellow passengers do not present the same danger, because adult riders help keep drivers alert and point out dangerous conditions and tend to talk less in heavy traffic or hazardous weather.

The research shows that having a conversation on a hands-free sets is as dangerous as a conversation on a handheld phone – the problem is that, unlike a passenger, the person on the phone doesn’t stop distracting you when road conditions change, and they aren’t a second pair of eyes compensating for your distraction.

So if the research is so strong, there are so many lives to be saved, how come we haven’t solved this one yet. Read on:

Joe Simitian, a state senator in California, managed to get his hands-free legislation, an effort he began in 2001, passed in 2006. He argued, based on data collected by the California Highway Patrol, that drivers using cellphones caused more fatalities than all the drivers distracted by eating, children, pets or personal hygiene.

In each previous year, the bill was killed — after lobbying by cellphone carriers, including Sprint, AT&T and T-Mobile. Mr. Simitian said that in the first two years, he would visit the offices of his colleagues on the Transportation Committee on the day of the vote and “find three cellphone industry lobbyists sitting in the legislator’s office,” Mr. Simitian said. “They’d just smile.”

• James Bennett: So you want a dynamic form
• Malcolm Tredinnick: Advanced Formset usage in Django

The steps are, create the group and add relevant users to it:

Change the project directory to be owned by that group, and accessible by no-one else:

Set the GUID bit on all the directories, so that new files and directories are created owned by the group:

Change the umask for everyone, so that new files are created with read and write permissions for the group:

The last part, changing the umask, isn’t ideal. It works on Debian and Ubuntu, because every user has their own group. I would rather a more focused solution, just for that directory – suggestions welcome.

References:

Mercurial and permissions
Multiple Committers
Change Ubuntu global umask
Collaboration models

…until internal temperature reaches 74C (165F).

How many office kitchens have a cook’s thermometer? Score nothing for usability.

Should you for any reason attempt to sue the manufacturer, it will rapidly become apparent that you didn’t follow the cooking instructions. Score one for legal.

Building tracks the wrong way

If you’re anything like I was, all your train layouts probably look like this:

You can only run one train on that track, but say you’re happy with that. When you need to connect another station, you might, unsuccessfully, try this:

Notice the three two-way signals. A signal locks an entire section of track from that signal until the next signal or the end of the line. These signals define four locks, color coded on this screenshot. If the train from Lundinghattan Ridge is in the Mardingbury station, it will have a lock on the yellow section, but not on the green section. The signal nearest Mardingbury will be red, but the other two signals will be green. The train from Marbourne will be able to acquire a lock on the green section, and stop at the signal nearest Mardingbury. We have a train stand-off. Not good.

To make that layout work, you’d need to remove the signal nearest Mardingbury, thereby merging the green and yellow sections. You remove that, and you have two trains sharing a station. OK, so now you add a third station to your network. Now things really start to break down.

The blue section is shared between the Lundinghattan train and the Chenningpool train. The Lundinghattan train is top left, just leaving Mardingbury station. It has the lock on the yellow section. Notice the two signals nearest it are red (actually all the signals in this picture are red, but focus on just those two). The train from Chenningpool acquired the lock on the blue section, but and this is the first important concept of this tutorial, once it got level with the depot it had a choice of two paths: Mardingbury, which is blocked by a red signal, and the depot, which isn’t. A train faced with a red two-way signal will always avoid that signal, even if that means going away from it’s destination. If instead of the depot we had a track running to the other side of the map, our Chenningpool would of happily headed down it, to avoid the red signal.

In practice this means our Chenningpool train will head into the depot, turn around, and head back to Chenningpool. It will never make it to Mardingbury. There is something very wrong with our approach, and the short answer is that we were using two-way tracks and two-way signals. We need to think one-way. Let’s start again.

The basic loading loop

Every shared station should have a one-way loading loop.

Notice the signals around the loop are all one-way. To place a one-way signal place a signal as normal, then click the signal again, once or twice depending on the orientation you want for your signal.

Now let’s connect our loop up to a town, and run two trains betweens those two towns.

We connected the shared track from Marbourne to our loading loop, with two short one-way sections. We can see the back of a one-way signal in the red circle, and the front of a one-way signal just to the right of the blue circle.

Let’s look at what’s going on in this picture. The train circled in red has the lock on the red section of track, and is held at the signal circled in red. It is waiting for a lock on the blue section of track. Notice that it could of kept going around the loop, instead of branching off and stopping at the red signal. Faced with a red one-way signal and a clear track going the wrong way, the train will stop at the signal, which is nearly always what we want. This is exactly the opposite to what would of happened with two-way signals.

The train circled in blue has the lock on the blue section of track, and is about to acquire the lock on the yellow section. As soon as it does, it will release the lock on the blue section, and the train circled in red will move forward. This is a layout that works.

Prefer one way tracks

Let’s connect up the other two towns, and not get blocked this time. The trick is to make all shared sections of track one-way.

The only two-way signals in this picture are circled in blue. All the others are one-way. The two-way signals are there to prevent a train on the two-way track from locking part of the one-way loop. If the left-hand two-way signal was not there, a train in Lundinghattan station would hold a lock on it’s two-way section of track, and the bottom part of the one-way section, up to the next signals. Remember, a lock is between two signals or the end of the track. Incidentally, stations don’t end a lock. If you had a station half-way along a track, the lock would run right through it until the next signal.

Pre-signals, the pro-layout

Mardingbury is getting quite busy now, we’d like to have two tracks in the station. Stop all the trains (or be quick!), bulldoze the station, and build a new, two track one. I moved mine back a square to allow space for the tracks to merge. and made the loading loop a little bigger. To control access to a multi-track station, you need pre-signals.

Pre-signals come in two types, entrance and exit. An entrance pre-signal will be red if all the exit pre-signals behind it are also red. The motivation for pre-signals is nicely illustrated here: Pre-signals on the OpenTTD wiki.

The entrance pre-signal is circled in blue. Notice that it has a horizontal white-bar, to show it is different. The exit pre-signals are circled in purple, and have vertical white bars. There is currently a train in the station, so one of the exit pre-signals is red. Because one of the tracks is free (green signal), the entrance pre-signal is green. The next arriving train will correctly go to the empty track. Even though they are pre-signals, we are still using one-way signals

What you can’t see on the picture, but which are very important, are the two normal one-way signals circled in black. They control station exit, by forcing a train wanting to leave the station to acquire a lock on the yellow section. This prevents two train leaving at the same time crashing into each other.

When a train is in the station, it still holds a lock on it’s section of track. The lock runs from the exit -pre-signal at the entrance to the station, to the regular one-way signal at the exit of the station.

Scaling it up

You now know all the key concepts, the rest is just more of the same. Here for example is what you would do if Lundinghattan got busy.

You give it a loading loop, and a multi-bay station. Pre-signals control station entrance, and regular one-ways control the exit. You can see the one-way’s at the exit much better on this station.

There’s one final change we need to make to allow lots of trains – we need to replace the two-way section highlighted in blue with two one-way sections.

We don’t have any more two-way signals. Each station has a loading loop, and one-way tracks connect the stations. In our first tries we had one track connecting the stations, and could only run one train between them. Now we have two tracks connection the stations, and in this picture alone there are eight trains, all serving Mardingbury. Now that’s more like it!

The stations are quite close together, so it might not be clear what is loading-loop and what is the tracks that connect them, so here’s an example with stations further apart.

Other stations would have their own loading loops, and as long as the one way tracks connect, you end up with a network spanning the world. Trains can run from anywhere to anywhere, and new stations just need plugging in to the network.

I have one final tip: Playing with virtual toy trains can be quite addictive, so remember to get some sleep :-)

Extremist material of any kind always looks gaudy and cheap, like a bad pizza menu. Not because they can’t afford decent computers – these days you can knock up a professional CD cover on a pay-as-you-go mobile – but because anyone who’s good at graphic design is likely to be a thoughtful, inquisitive sort by nature. And thoughtful, inquisitive sorts tend to think fascism is a bit shit, to be honest. If the BNP really were the greatest British party, they’d have the greatest British designer working for them – Jonathan Ive, perhaps, the man who designed the iPod. But they don’t. They’ve got someone who tries to stab your eyes out with primary colours.

Read the article: Charlie Brooker on the BNP and their political broadcast.

My talk will be entitled How and Why to Extend Firefox in Javascript (and Thunderbird, Komodo, and Songbird). I will post the slides here in June.

See you there!

The game had a splash screen of a Far Side comic, then a short video of me tromping through the woods like a Neanderthal, which my Dad filmed and which I digitized with a very early video capture card.

In Borland’s Turbo C++ 3.0 I wrote a basic graphics engine to display the tiles of the world, and an event loop so I could move the main character around the world. I drew sprites for a proto-human (the micro zooid), dirt, rocks and sticks. He could walk around the world, and pick up and put down rocks or sticks.

Then I took a break to plan. I have a proto-human, rocks, and sticks. How do I get to civilization?

Edit /etc/subversion/servers
Add this line at the end: neon-debug-mask = 511

That showed me this error: auth: '/' is inside auth domain: 0.

This means that the path I was requesting (the root of the repo) was not considered inside the AuthDigestDomain I had set in Apache.

It turns out that at some point in the upgrade of Apache, Subversion, or a library, the AuthDigestDomain requires a scheme. I had
AuthDigestDomain svn.myserver.com
whereas it should of been
AuthDigestDomain http://svn.gkgk.org.

So now you know.

Most of what Picasa does can be replicated with the Colors / Levels or Colors / Curves tool. It’s well worth spending a little time experimenting with both of those (the documentation is very good too).

Crop

In the Toolbox, click the Rectangle select tool
In its options (beneath the tools), tick ‘Fixed: Aspect Ratio’
Enter 6:4 ratio (for 1.6 sensor, most DSLRs)
Tick Highlight. Draw a rectangle on the image that you want to crop to.
Image menu / Crop to Selection

Straighten

Click one of the rulers above or to the left of the image, and drag a guideline onto your picture
In the Toolbox, select the rotate tool
Select ‘Clipping: Crop to result’. Maybe ‘Interpolation: Sinc (Lanczos3)’, although that doesn’t seem to matter
Rotate until your picture is straight, using your guideline
Click Rotate
Image / Fit Canvas to Layers or Image / Autocrop Image
Image / Guides / Remove all guides

Redeye

Filters / Enhance / Red-Eye Removal (I have never used this)

I’m feeling lucky

Colors / Levels / Auto

Auto Contrast / Auto Color

Colors / Auto / something

Fill light

Colors / Levels
Drag the middle triangle (grey) to the left

Highlights

Colors / Levels
Drag the right side (white) triangle

Shadows

Colors / Levels
Drag the left side (black) triangle

Color Temperature

Tools / GEGL Operation / color-temperature
Adjust Intended Temperature

Neutral Color Picker

Colors / Levels
There are three color pickers near the bottom right
Use the left one to select black, the middle one neutral gray, and the right one white

Sharpen

Filters / Enhance / Unsharp mask
Try these values: Radius: 1 – 5 Amount: 0.5 – 1
OR
Colors / Components / Decompose.
HSV, Decompose to layers
Switch off the hue and saturation layer
Apply the Unsharp mask, as detailed above
Colors / Components / Recompose

Sepia

Filters / Decor / Old Photo

B & W

Image / Mode / Grayscale
OR
Colors / Desaturate

Warmify

Colors / Curves
Select Blue – pull the center-right of the curve down most of a grid box
Select Red – pull the center-right of the curve up most of a grid box
OR Tools / GEGL Operation / color-temperature / Increase intended temperature by 10k or 20k

Saturation

Colors / Hue-Saturation / Pull the Saturation slider to the right

Soft Focus

Duplicate layer (right click / Duplicate or use the icon bottom of layers pane)
Filters / Blur / Gaussian Blur.. Set to 60
Reduce Opacity to ~60%
Right click on blur layer, Add Layer Mask, White (full opacity)
Click the foreground color, set S to 0 and V to 50 (or 60, 70)
Select a brush, the Paintbrush tool, paint over the parts you don’t want fuzzy
To replicate Picasa this would be a big circle somewhere in the middle of the picture

Graduated tint

Duplicate layer
Switch off Background by clicking the eye
Edit that layer with Levels and Curves to expose sky correctly
Right click on new layer / Add Layer Mask/ White (full opacity)
Select Blend tool
Draw a line on the image to make a gradient. Try again.
Click the Background eye back on
Right click on the edited layer, and Apply Layer Mask
Merge the layers

If that sounds familiar, you’re not alone. There’s:

• a project to find uses for LinkedIn
• a book on finding a point to LinkedIn
• and the company itself had to blog about ten ways to use LinkedIn.

If it takes a blog post, a book, and a community project, to find a point to your web application, I think there may not be one.

LinkedIn tries to explain itself

Take the blog post for example. It says that LinkedIn is, and I quote, “a great way for professionals to strengthen their online brand reputation and leverage their professional network”. Huh?

How about the front page. I bet they have a super-succinct value proposition. It’ll be obvious once I get to the front page:

• Stay informed about your contacts and industry
• Find the people & knowledge you need to achieve your goals
• Control your professional identity online

Let’s be honest, the reason you are ‘connected’ to these people on LinkedIn, as opposed to Facebook, Bebo, Orkut, or Friendster, is because you don’t want to “stay informed” about them. You don’t want to know what movie they watched last night.

What about the second bullet point, finding people? That’s about recruitment, and the trouble is, recruitment has gone niche.

• If I need a Python programmer, I post on python.org,
• for Django I post on Django Gigs,
• front enders are on Authentic Jobs,
• and so on.

LinkedIn seems to be built around a 1940’s model where you hire someone because they went to the same college as you: “Cambridge lad eh? Jolly good. Welcome to the company. Scotch?”

Maybe it’s all about that last bullet point, your “professional online identity”? LinkedIn is your home on the web, a virtual calling card, MySpace without the teenage exuberance? Well, if you’re in tech and your only online presence is LinkedIn, that’s less than impressive. A bit like a graphic designer printing his business cards on that machine at the mall.

LinkedIn prevents people from contacting you

If you are trying to get in touch with someone, whom we’ll call Chris, you probably start with your favorite search engine. You type in his name. Chris’ LinkedIn profile comes up. You click on it. You now have three options:

• You can pay LinkedIn $25, and they’ll let you send Chris an email (an ‘InMail’).
• You ask your friend Alice to ask her friend Bob to introduce you to Chris.
• You do what I do. You hit Back. Click on the next search result, which is Chris’ blog, select the Contact or About page, and email him directly. LinkedIn just wasted your time.

I may of mentioned this before, but this is not the 1940’s. We Internet types are comfortable talking to strangers. All I want is an email address, and that’s what LinkedIn doesn’t give me.

Let’s recap

You go to LinkedIn. You declare all your connections. You complete your profile, recommend people, answer questions, and so on. You give them a lot. And what does LinkedIn do for you? They prevent people contacting you. So, what, exactly, is the point of LinkedIn?

Foxden is a Firefox extension that allows you to tile all the web applications you use on one page. Imagine being able to see your email, your calendar, your bug tracker, feeds, twitter, whatever you use, on one page. Take a look at my setup:

As you can see, I have (counter clockwise from top left) my email, calendar, feed reader, todo list and a local text file for taking notes.

It’s free, should work wherever Firefox 3 works, and it could be yours right now.

Download the Foxden Firefox extension

Instead of databases becoming available as a result of Freedom Of Information Act requests, government officials should be required to justify why any public data should not be freely available to the taxpayers who paid for its creation.

Wow, what an exciting time to be in North America.

From the O’Reilly Radar.