Henri Bergius - Desktop

Working on an Android tablet: first six weeks

henri.bergius@iki.fi (Henri Bergius) — Tue, 16 Apr 2013 00:00:00 -0700

I've been working full time on my Android workstation for over a month now, and it is time to write an update about it. How has it worked out?

What I've been doing

I would love to tell stories of working from parks and cafes, like Mark O'Connor has on his iPad setup, but unfortunately we had a backlash of winter here in Berlin and the warm spring weather only came back this week.

Instead — quite atypically — I've been mostly desk-bound in this time. The EU projects that mandated a lot of travel have now ended, and my current projects are more about software development than evangelism.

But that actually makes this experiment even more useful, as it means most of the six week has been actual programming, which is what most of my readers also do.

For those who missed my setup in the previous post, this is how it looks in action:

On-screen are tmux, vim, and a Grunt test automation watcher running inside a MOSH client.

Here are some things I've done in the last month:

Porting the NoFlo flow-based programming engine to run in both browser and Node.js with the same codebase, including a tutorial on how others can do the same
Writing and publishing an implementation of Android-style Action Bars for web apps
Adding multiple major features to the web-based NoFlo IDE
Dealing with the issues raised with the release of jQuery UI 1.10 and Backbone 1.0.0 with Create.js and VIE
Blogging, including publishing some quite popular posts

In general the experience has been a positive and productive one. I'll write about some nuances here.

Web debugging

As you can see from the list above, much of my recent work has been client-side. With this, the unavailability of web debuggers on mobile browsers can become a problem.

I've tackled this issue in two ways:

More tests. Instead of poking around in a debugger, I try to write Mocha tests for most aspects of my applications. This also has the benefit of automation, meaning that PhantomJS will test everything in my application every time I commit
VNC and desktop browsers. When I really need one, I can still use the web debugging tools of traditional web browsers via VNC

See Mark O'Connor's setup instructions for VNC on one of these tablet workstation setups.

Post-PC means post-Office

One area where tablets are really lacking is support for traditional office tools like word processors and spreadsheets. There is a Google Drive client, but it is very slow (even small spreadsheets can take minutes to load) and mostly non-functional (word processor doesn't even support headlines).

There are also some other office suites available, but even these are better used for viewing documents than actually making changes to them.

But the bigger question is whether traditional office tools even have a place in this modern world. The commentary on constant delays with Microsoft Office for iOS and Android shows that people don't see them as that relevant any longer:

For the longest time, Office was the ubiquitous productivity suite. Everybody used it. Nobody considered using anything else. However, since this mobile revolution started, even non-geeks are starting to question whether Office is still all that. I had breakfast this morning with a CPA who does all of his work in Google docs. There is an entire generation of future workers going through high school and college now who don’t even have Office installed on their computers. If Microsoft has any hopes of keeping Office relevant, it needs to be everywhere, including the iPad.

Personally, I might be a lot better off writing my documents in Markdown, versioning them with git, and maybe using custom data-gathering applications with CouchDB map-reduces for data visualization.

The story of tablet productivity is still evolving. The new tools and interaction techniques we have will eventually give rise to new kinds of productivity applications. That may signal the end of the Office hegemony.

Presenting from the tablet

On the first week of this experiment I was actually traveling. The final review meetings for both of the EU projects were being held in Brussels and Luxembourg, and I had to present our results.

As the presentation tools on Android are not very good, I took this as an opportunity to finally start using an HTML-based presentation system. For this, I picked DZSlides, with a custom Jekyll-based flow for constructing slide decks from individual assets stored in git.

The results are quite nice, and I love being able to embed live demos inside the slides.

With every computing platform, there is always some fiddling involved with getting your device connected to a beamer. I was positively surprised with how easily the Nexus 10 worked. Simply connect using a micro-HDMI to VGA adapter, and you'll have the tablet screen up on the projector.

Minor annoyances

Everybody knows about the common gripes with Android on large tablets — most apps have been written with a narrow phone screen in mind, and simply look bad on a wide 10" screen. Twitter is a good example of the typical neglect of Android's UI guidelines.

Somewhat surprisingly, this even applies to Google's own tablet applications. Apps like Google+ and Google Drive are a lot more functional on an iPad than on a large Android tablet.

However, these are more of a problem when using something like my Nexus 10 as a media tablet, and don't really affect how well it works as a programming workstation.

For programming work, what matters is things like the beautiful screen, Android's reasonably good support for hardware keyboards, user-accessible file system, and the ability to share information between applications. These are the main reasons why I went with Android instead of an iPad.

However, not all is sunshine. So far, the main annoyances for me have been:

Regressions in Magic Trackpad support mean that it is practically unusable when you also have a Bluetooth keyboard. A lot of character presses get duplicated, making typing near-impossible. I'm assuming other mouse devices would however work
MOSH ConnectBot — which I'm using for my programming sessions — makes Ctrl and Esc keys not work. Luckily I was able to find a workaround
Android 4.2.2 is buggy on the Nexus 10. Especially Chrome can cause the system to crash. Other browsers help here, and hopefully Google will fix the issue soon

Now, crashes and freezes may sound like a big deal. But thanks to using tmux, they just mean a short interruption, and not any lost work. I just restart my tablet or MOSH client, attach back to the tmux session I was working with, and I'm right back to where I was, cursor position, vim splits, and all.

Quantifying productivity

Calculating programming productivity is notoriusly difficult. While Mark was able to show impressive figures from his iPad setup, I don't have anything similar because:

I haven't had the time to crunch the numbers on the work I do
The ending of the EU projects meant that I'm now doing different things than I did with my laptop, and so comparing results from the two is hard

And in the end what matters more is the results of the work, not the effort spent getting there.

But still, it would be good to have a bit more data on how well the setup works besides the subjective "it feels like a good way to program".

To this end, I recently started using the RescueTime tracker on both of my Android devices. It keeps calculates how much time I spend with different applications each day, and even allows me to give some sort of productivity scores for them:

I'll be running this for a while, and will try to combine it with some statistics from my GitHub account. Those two should be able to paint a picture of how I work.

Conclusions

In the beginning, like with any new tool you have to start using, the Android tablet setup felt weird and limiting. But it has grown on me since, and right now I'm not regretting giving my laptop away.

But at the same time, if a new interesting device came out, the cost of switching to that would be minimal. After all, the Nexus 10 for me is essentially just a window into the web and my terminal running somewhere else.

In a way this decoupling is similar to the traditional desktop PC setup where you have a separate computer, screen, mouse, and a keyboard. The difference here is that none of those parts are bound to a desk or connected with cables. Instead, the peripherals talk with my screen over Bluetooth, and my screen with the "computer" over the Internet.

If I for instance want a better keyboard, I can just buy one and replace that part without having to change anything else with my setup.

Cost advantage

One aspect that people have remarked on is cost. Over the course of two years — which is the typical replacement cycle of a professional workstation — this setup is cheaper than a MacBook Air. And with that price I get a lot better screen and about double battery life, and even a smartphone.

What I lose is the ability to work fully offline, though somewhat alleviated by having local vim and git via Terminal IDE.

Moving forward

The experiment will keep continue. In these six weeks, I haven't seen any negative impact on my productivity from working on an Android tablet instead of a laptop, and many positive ones. Portability, battery life, and the emphasis on tests and automation are probably the foremost.

Of course, new devices come to the market, and eventually something will come that beats the current setup. But then I'll be able to switch without even losing my cursor position, so the only cost is the hardware itself.

In time, I will write more about how things are going.

Working on an Android tablet

henri.bergius@iki.fi (Henri Bergius) — Mon, 18 Mar 2013 18:00:00 -0700

As mentioned in my post Hacker-nomad's toolkit, 2012 edition, the lease period of my lovely — Linux-driven — 11" MacBook Air expired this month, and I had to consider what kind of gear to go with next.

The safe bet would've been to just get a newer version of the Air, or maybe the 13" Retina MacBook with its great screen. A fresher approach would be a ChromeBook, either the cheap and light ARM ChromeBook, or the Pixel with awesome screen and design but crappy battery life.

I'm spending most of my days developing software, and so I should be able to work with a ChromeBook and a remote Linux box. But if that works, why not try working on a tablet? They're cheap, light, durable, and have an all-day battery life. And if Mark O'Connor was able to work productively a whole year with one, why couldn't I?

Since I already had a Nexus 10, this is what I decided to try.

The setup

The Nexus tablet has a great, "better than retina" screen, which can render my coding sessions and web user interfaces beautifully. The 10" screen is somewhat smaller than what I had on my Air, but not terribly so.

However, the big problem with tablet productivity is text input. Much of our current work environment is still textual, and typing on a non-haptic glass touch-screen is simply not very nice.

Hardware keyboard

To fix this I purchased a Microsoft Wedge Keyboard, which connects to the tablet over Bluetooth. Recent Android versions have quite full support for external keyboards, allowing me to use it for all text entry, and even for some keyboard navigation. So yes, Alt-Tab works.

The Wedge keyboard is about the same size as typical keyboards on compact laptops. Microsoft has always made good hardware, and the keyboard is no exception, providing a quite nice feel for the size it has. A handy additional feature is the included cover, which puts the keyboard to sleep automatically, and can act as a tablet stand when opened. No more issues with keyboard waking up in your bag and deleting everything.

For those prepared to lug a heavier option, there is a Bluetooth mechanical keyboard, which is probably a lot better than any laptop keyboard on the market. And if you already have a good keyboard, Android supports most of the USB ones via an OTG cable.

Desk setup

While I'm spending quite a lot of my time on the road, living out of my backpack, I do have a regular desk in the Berlin office I share with Content Control. Since my coding sessions are often long, I've been a bit concerned with my programming ergonomics for a while now, even considering a standing desk.

Tablet has the same advantage as a traditional display in that it is decoupled from the input devices, giving you greater freedom in how to position them. I have a Callstel tablet stand that allows me to place the tablet in practically any place and height above my desk. The current setup is just slightly below my eye height in the normal sitting position, but I'm still experimenting with that.

Gorilla arms

The Gorilla arm syndrome is what everybody brings up with every touchscreen computer — it is simply not nice to constantly lift your arm to touch the screen.

In my experience this isn't so much of an issue when you're using the tablet positioned similarly as a laptop screen would. But when the screen is up in a more ergonomic position, like it is on my desk, then this quickly becomes an issue.

To solve this I bought a Apple Magic Trackpad, which connects to the tablet again via Bluetooth, and allows both regular mouse usage with Android, as well as many multitouch gestures.

Why Android

Most people experimenting with replacing computers with tablets go with an iPad, the established market leader. iPad has many benefits over the Nexus 10, including a more mature software ecosystem, and better availability — if you break or lose your tablet, you'll be able to pick up a new one from practically anywhere, whereas the Nexus devices are only available online.

The reasons for me to go with Android have to do with openness. Since the core operating system is open source, there are custom ROMs I could use if I wanted, and I can do file management the traditional way when I need to. An even bigger reason is that the sharing system makes it possible to connect various applications together. Being able to run multiple broser engines is also nice for a web developer.

Another bonus is the availability of NFC on the tablet. I'm quite often sharing content between it and my smartphone. If I run into an interesting web article, I can either send it to my Kindle to read later, or touch the tablet with my phone and read the article on that. This is surely a feature that will gain more mindshare whenever it is introduced to iPhones.

I also prefer the rugged, rubberized look-and-feel of the Nexus 10 to the cold metallic iPad, even though a 4:3 screen would be better than the widescreen I have now.

Software used

I really don't need much for my daily work — just a browser and a terminal. Here are the apps I use on a daily basis:

Chrome and Firefox web browsers
JuiceSSH SSH client to access my remote Linux server
TapChat IRC client
File Manager HD for moving stuff around, including between file servers
DroidEdit Pro for quick local file edits

In addition I'm using some of the built-in Google apps, like Google Talk, GMail, and Google+ for Hangouts.

For offline development I have an installation of Terminal IDE that allows me to run Linux utilities like vim and git locally. If I would root my tablet I could also install a Ubuntu chroot and run whatever I need. With a previous tablet I even was able to run Node.js servers and databases on the thing!

My development virtual machine is from DigitalOcean's Amsterdam site, providing quite nice fast connections here in Europe. I mostly work on it via tmux and vim, and run whatever processes I need, including long-running NoFlo flows.

Pros and cons

Based on my initial experiences of working with this setup for a week, working on a tablet is quite different from a traditional computer. Here are some good things:

Battery life: the Nexus can easily get me through a full workday with a single charge, meaning that I only need to connect it to a wall overnight. It also charges via standard micro-USB meaning that I don't need any extra power bricks with me
Portability: most tablets are smaller and lighter than full laptops. And they can be used more easily when standing, sitting on the couch, etc.
Instant on: there is no suspend/resume cycle. I press the power button, the tablet recognizes my face via the camera, and I'm instantly back to where I left off
Modularity: with a tablet, my work environment is built out of multiple modular pieces that I can take with me, or leave at the office depending what I intend to do. And I can set them up in different configurations when working
Touch: for software developers, working on a tablet really drives home the importance of touchscreen friendliness. I've already noticed this affecting my UI designs
Focus: everything is full screen, meaning no need for window management. Tablet software also tends to be simpler and has less configuration to fiddle with

But obviously there are some downsides as well:

Office documents: the office suites available for Android are quite poor, and the mobile version of Google Docs is simply terrible. One solution would be using MS Office or LibreOffice over a VNC connection
Mobile-first web: quite a few websites try to offer even large tablets like the Nexus 10 their silly mobile sites. Thankfully this is becoming less prevalent due to media queries and responsive design
Offline: much of my current tablet workflow requires me to be online. I could write code and blog posts offline with tools like Terminal IDE, but there would be no way to run and test software
Bugs: many tablet applications are still in their first generation and lack the maturity and robustness that their desktop counterparts have had time to gain
Web debugging: while WEINRE sort of helps here, it is still a lot less convenient than the web development tools that come with desktop browsers

Conclusion

This tablet work setup is for now an experiment. If I find it hindering my productivity, I'll just have to get one of the laptops mentioned in the beginning of this post and work in more traditional manner. But if it works, then great! In that case I have finally found a more modern setup for programming work — one that gives me both better ergonomics and mobility.

I will try this setup for some period of time, and then report the results here on my blog.

Building a smarter workplace

henri.bergius@iki.fi (Henri Bergius) — Fri, 08 Mar 2013 00:00:00 -0800

As part of the SmarcoS project, we have been investigating how to make workplaces smarter through sensors and context awareness. Here is a video showing what we've built:

The idea here is to facilitate collaboration and smoother project communications through various different tools that I'll describe below. While this already does a lot, it is obviously only the first step on the path to making offices smarter.

Office presence

An important part of collaboration is to know who is where. Maybe some people are having a lunch break, or are working remotely? The Office Presence Display system knows these things, thanks to various sensors:

Bluetooth sensors can see smartphones and other mobile devices in the space
WiFi sensor can see what computers are connected
Google Talk sensor knows who are connected and active on their work accounts

With these, we know pretty well when you arrive to the office, and when you leave. Thanks to the sensor watching the company instant messaging system, we also know when people working remotely are available.

In addition to letting people know where the other members of the team are, this system can help with the perpetual annoyance in many companies: having to fill timesheets. We have a logger process that listens to the sensors, and logs the data into a CouchDB database. From there you can easily visualize working hours and availability trends:

In many situations it is of course not enough to know whether people are present, but also to know what they're working on. The next component of the system helps with that.

Electronic Kanban wall

Instead of the clumsy waterfall style, more and more companies are managing their projects in tight, recurring iterations. A Kanban wall is a great way to keep track of tasks as they move through the process, and to see who is doing what, and what could be possible to do next. The One day in Kanban Land post explains the concept well.

If everybody working on a project sits in the same room, and that room is also where all decisions about the project are made, then the traditional solution of whiteboards and Post-It notes is probably the best way to visualize Kanban. But for distributed teams and more flexible work, an electronic version is a lot better option. This way everybody can see the Kanban wall in its current state from their computer, a TV in an office room, or a tablet.

Our Kanban wall implementation provides exactly that. The same wall (or, in case of multiple projects or teams, a set of walls) is available through any web browser. You can also display it on a big screen in an office or a meeting room, and control that screen with the Kinect Air Cursor.

Now, for most companies this is not meant to be the master database of projects and tasks, but instead just to be an alternative view and controlling tool to wherever your project information resides. Maybe you're already using Pivotal Tracker, Basecamp, or OpenPSA? The Kanban wall server has a full REST API available, and so it is easy to integrate with any existing system. If a task is moved on the Kanban wall, it can be updated to the project tracking system, and vice versa.

If you're using both systems, then the Kanban wall is also aware of the presence information. We show a differently colored border around the "person tokens" depending on the availability state. You can also set different Work-in-Progress limits for different people and states.

Interested yet?

So far we've only trialed the system in our own office, but are now looking for some pilot customers to try the system out. If you're interested in making your workplace smarter, get in touch.

For those who like to tinker with these things, all the sensor software is available as open source at http://github.com/nemein.

Hacker-Nomad's toolkit, the 2012 edition

henri.bergius@iki.fi (Henri Bergius) — Fri, 21 Dec 2012 00:00:00 -0800

Back in last July -- when choosing photos to use in the epic The Dreams of the MeeGo Diaspora post -- I noticed that it is quite fascinating to look at the various tablets and mobile phones we've been using over the years. Back then they all were so new, shiny, and exciting, and yet hardware moves so fast that something from a year or two ago would look quite dated now.

And so I thought it might be interesting to keep some sort of record on what kind of computing devices I've been using for my work -- and how -- over the years. This post is the first of the series, and shows the setup I've had during 2012.

The setup, and what I do with it

Many programmers believe in beefy workstations and huge multi-display setups. I was never so much into that, and as I'm spending half of my working time on the road, it has made sense to adapt to a more minimalist setup. A small laptop, a small tablet, and not much else.

Most of my programming is for web services, and so the main things I need are a browser and a terminal. After the move to Berlin, various video conferences with project partners all around the world via Google Hangouts or Skype are also almost a daily fare. My current setup handles these use cases quite well, and is simple and light to be carried around Europe without any problems.

At the same time, the nomadistic lifestyle has meant there is really quite little use for a phone. I do couple of calls or SMS per week, but in general the roaming charges, and better communication tools that Internet offers means that phone is just another screen to the services I use, essentially just a small tablet.

Workstation: 2010 MacBook Air 11"

My current main computer is the MacBook Air I got in early 2011. Apart from the lousy battery life, I've been generally very happy with it. It is small, light, and still has a reasonably good screen and keyboard. The solid state drive makes it also quite fast.

I'm running this laptop with Ubuntu 12.04 and the GNOME Shell. Linux supports this hardware quite well, meaning that I've never had issues with things like sound, the multi-touch trackpad, WiFi, or suspending. But external displays have sometimes been tricky.

During a typical workday, I only have two windows open: a full-screen Firefox window for email, browsing, and testing the software I write, and a full-screen terminal window running Vim with various splits. All of the rest of the software runs usually on a cloud hosting provider, or inside a Vagrant virtual machine. I sometimes also run Libreoffice, but more and more of my office productivity needs are now handled by Google Drive.

The battery life issue is what's making me rethink this part of the setup. Apple promises a new MacBook Air about five hours of use, but after two years, and with running Linux instead of OS X, I'm down to maybe two and half hours. This is by far not enough, especially during conferences or travel where power outlets are not readily available.

I'd also love to get to a device with a bit more standardized ports. Having to replace faulty Apple laptop chargers and to remember the display dongles when presenting something is a bother.

Tablet: Nexus 7

Having used tablets daily for a long time, I'm a big believer in them as both production and communication devices and as a much better computer during free time. Actually, to help maintain a sensible work-life balance, our household of two programmers has a strict no laptops at home policy.

I used a Transformer Prime convertible tablet during the first half of 2012. With its detachable keyboard it was a great device as a travel computer and for conference live-blogging. The keydock contained an additional battery, and so I could rely on being able to use the device for a full day.

At the same time, the tablet was a bit big and clunky for travel. You'd always have to dig it out of the bag separately, and such a big device wasn't comfortable when reading in bed. And so, when the Google's Nexus 7 came out, I swapped devices. The seven inch form factor is great for both reading at home, and during travels. When you need to swich trains or board an airplane, you can just stick the device in your pocket. It certainly isn't comfortable to keep there for long times, but pn short transfers this is a very handy possibility. No wonder people on the iOS side of the fence are so excited about the iPad Mini!

I use the tablet for the various video conferences I have, and do most of my web reading with Firefox Mobile and Instapaper. On work trips I also read books on the Nexus, so that I don't need to bring the dedicated Kindle e-reader as an extra device. Firefox Mobile is great in that it synchronizes my passwords and browsing history between the tablet and my laptop, so that I can always easily pick up and continue on either device.

Internet access: Huawei MiFi

Since much of what I do happens online, being able to connect from everywhere is very important. While I could use JoikuSpot to share the Internet connection from my Symbian phone, having a dedicated device for this purpose makes sense. With the MiFi wireless base station I can spare the battery of my phone.

As the MiFi is only used for Internet access, I can buy cheap pre-paid SIMs from each country I travel to. Paying somewhere around ten euros for a month of Internet abroad certainly beats the usual roaming charges!

However, having to dig this device from my bag and start it whenever I need connectivity is a slight hassle. Now that there is a Nexus 7 with 3G support available that might be a better option.

Audio: Sennheiser noise-cancelling headphones

This is actually a piece of gear I've had for a while: For music and teleconferences I'm using Sennheiser's noise cancelling headphones that I bought back in 2008. While they're bulkier and pricier than typical in-ear headphones, they are great especially when traveling or working in a noisy environment.

The headphones are slowly getting a bit frayed, and so I will probably have to look for replacement soon. Having optional Bluetooth could be handy.

Thoughts for 2013

The current setup has generally served me well for about two years of intense travel and open source development. We generally run our computers with a two year replacement cycle, and so at least my laptop is due to be replaced quite shortly.

Thanks to services like Travis CI and Heroku, my computer needs are becoming even simpler than they used to be. Theoretically I could work with any device that has a good screen, a good keyboard, and can run terminal and a web browser. This means I could even work on an iPad. At least nearly.

For the next cycle of devices, the focus will be on features like:

Endurance: the battery should last a full working day when traveling or at conferences
Connectivity: I want Internet everywhere, without any unnecessary hassles
Resolution: having used some of the high-DPI devices like the new iPads and MacBook Pros, I definitely want something like that

I'm also intrigued about switching to a Kindle Paperwhite 3G as my reading device. Better screen, and ability to access WikiPedia anywhere in the world for free!

If you have any great suggestions for a device or a combo fulfilling those requirements, please comment!

If this all sounds very consumeristic, it should be noted that my old work devices always get recycled to other users. When you do mobile and web development, you need to have access to the new hardware that people are buying and using.

Midgard 12.09 Gjallarhorn

henri.bergius@iki.fi (Henri Bergius) — Tue, 02 Oct 2012 13:49:00 -0700

The new stable series of the Midgard2 Content Repository library was released recently. This version builds on the long-term supported Ratatoskr series, adding some new features:

Asynchronous I/O operations with the content repository
Easier migration from the Midgard1 series

Asynchronous I/O is important when the content repository is used in persistent applications like desktop software or Node.js, where Midgard can be used via node-gir (see examples). You can read more about async operations in the documentation.

Why a content repository?

From the release notes:

Midgard2 is a library which provides content storage and retrieval services to applications. It is essentially a higher-level access layer to relational databases and file systems.

Parallels can be drawn between Midgard2 and various Object Relational Mapping (ORM) libraries. The content repository concept however takes these ideas much further, with concepts like:

Object-oriented data and query access

Tree structure for content

Standardized metadata available to all content

Workspaces for managing branching and merging of content

Content type definitions with introspection capabilities

File attachments for content objects

Signals about I/O operations

Users and access control

On desktop and mobile

Midgard2 is available through GObject Introspection, making it accessible from a wide range of programming languages, from Python and JavaScript to C and C#.

Simply install the Midgard2 library and the GObject Introspection binding for the language you want to work with.

With PHP

There are two ways to use the Midgard2 content repository with PHP:

Using the Midgard2 PHP extension directly
Via the standard PHPCR interfaces

For those looking for a CMS running on Midgard2, both MidCOM and Symfony CMF work with it out-of-the-box.

Getting Midgard2

It will take some time before the new release trickles down to distributions like Debian and Ubuntu. In the meanwhile, source tarballs are available.

If you have any issues getting started, feel free to contact us on the mailing list or on IRC.

Gjallarhorn of the Viking mythology is the horn that sounds marking Ragnaroek. With our release it signifies the callback pattern of asynchronous I/O, and the time for Midgard1 users to migrate over. The picture is public domain from Wikimedia Commons.

GObject Introspection and Node.js

henri.bergius@iki.fi (Henri Bergius) — Wed, 25 Jul 2012 00:00:00 -0700

Unfortunately I will not make it to GUADEC this year. However, here is something new for GNOME developers:

I wrote last year how GObject Introspection was coming to Node.js. Back then the API was still quite bad, and it was limited in what you could do with it. Since then things have moved forward quite a bit, and today Piotras released version 0.1.0 of node-gir.

This allows you to interface with any GObject Introspection capable libraries, including the ones you write yourself.

There are still some things to be done, especially with stability, but with node-gir Node.js essentially becomes a fully qualified GNOME development environment.

For example, here is a CoffeeScript port of the GNOME Guitar Tuner example. Still a bit crashy, but at least it shows how the node-gir API works:

#!/usr/bin/env coffee
# Guitar Tuner
# ------------
#
# This is a CoffeeScript and node-gir port of GNOME's guitar tuner example from
# <http://developer.gnome.org/gnome-devel-demos/unstable/guitar-tuner.js.html.en>
gir = require 'gir'
gtk = gir.load 'Gtk', '3.0'
gst = gir.load 'Gst', '0.10'

gtk.init 0
gst.init 0

guitarwindow = new gtk.Window
  type: gtk.WindowType.toplevel
  title: "Node.js Guitar Tuner"

guitarwindow.on 'destroy', ->
  gtk.mainQuit()
  process.exit()

guitar_box = new gtk.ButtonBox

playSound = (frequency) ->
  console.log frequency
  pipeline = new gst.Pipeline
    name: 'note'
  source = new gst.ElementFactory.make "audiotestsrc", "source"
  sink = new gst.ElementFactory.make "autoaudiosink", "output"
  source.set_property "freq", frequency
  pipeline.add source
  pipeline.add sink
  source.link sink
  pipeline.set_state gst.State.PLAYING

addButton = (tune, freq) ->
  button = new gtk.Button
    label: tune
  guitar_box.add button

  button.on 'clicked', ->
    playSound freq

tunes =
  E: 369.23
  A: 440
  D: 587.33
  G: 783.99
  B: 987.77
  e: 1318.5

addButton tune, freq for tune, freq of tunes
guitarwindow.add guitar_box
guitar_box.show_all()

guitarwindow.show()

gtk.main()

People interested in this effort should watch Piotras' node-gir repo. Now the project even has continuous integration on Travis using the Midgard GI bindings to test things.

Of course you can do other cool stuff, like creating a web browser with just couple of lines of JavaScript. Find more from the examples folder.

The Dreams of the MeeGo Diaspora

henri.bergius@iki.fi (Henri Bergius) — Sat, 07 Jul 2012 00:00:00 -0700

Much has been written about the emerging Post-PC era, about the new possibilities it brings, and the limitations it imposes on developer creativity.

For a while I've been planning to record my own experiences of the mobile revolution through the lense of the Maemo mobile Linux ecosystem. This is much in the same spirit as Texrat's excellent piece. Today's news made this even more urgent, and so here it is.

New Hope, a Linux computer in every pocket

For starters it ought to be said that I'm not a newcomer to the mobile Internet. Back in 1998 I was already blogging on an Internet-connected Psion PDA, and by early 2000s we were routinely publishing our travel journals this way. But even after the Psion experiences, Maemo was something special.

Glimpse of a tablet future

The real mobile story for me started only with Nokia's Internet Tablets and Maemo in 2006:

I had the Nokia 770 web browser device waiting on my desk as I returned from South Africa. 770 is a nice, small, Linux-powered internet appliance that is able to utilize either a WLAN connection, or mobile phone's connection via Bluetooth.

The device is compact and lightweight enough to be basically carried all the time. It also has a screen good enough for reading almost any websites. These two factors serve to make web and RSS feeds ubiquitously available, opening interesting possibilities

No wonder we felt it to be the real-world Hitchhiker's Guide to the Galaxy:

However, Douglas Adams didn’t probably have in mind what would be really possible today, a joke that was written back then is now reality. What is 770 + Internet + Wikipedia? Quite much same as Hitchhiker’s Guide to Galaxy. An electronic device that can answer to all your questions anywhere anytime. It is not Sci-Fi anylonger, but Nokia 770 Internet Tablet...

Many of the things people associate with iPad were already common for us in the old Internet Tablet times. I was getting my morning news on the 770 with Google Reader just like I now do with Pulse on an Android tablet, and I was sharing my location with friends via Plazes like people now do with Foursquare. The only difference is that back then the tablets were for a bit more exclusive club of Linux enthusiasts.

Community jumps in

The early stages of Maemo were important learning experiences for how a big mobile player could interact with the free software community. While not everything went perfectly, the 770 very quickly gained an active community of developers around it.

Next software updates to the device brought VoIP and instant messaging to the device, and the community brought mobile maps (later on Maemo Mapper would go with the more legit option of OpenStreetMap). The instant messaging capability proved itself useful very quickly:

We spent the last two days driving from Helsinki to Poznan, Poland for the Midgard Developer Meeting. This proved to be a good field test for Maemo 2.0 as we needed to instruct people back home about some project details using Google Talk.

The media capabilities on the 770 also made it possible to watch movies on the tablet.

Helping to run the community

Until this point we had been just regular community members, using our tablets for various purposes, and even writing some simple pieces of software for it. But in late 2006 our involvement in the community deepened extensively, as our Midgard system was selected for running much of the community infrastructure:

We have conducted a study on migrating/integrating all existing services of maemo.org with the Midgard CMS framework. Based on the study we have decided to go ahead and setup the new environment ... Nemein is participating in the project as the Midgard expert organization.

Those interested in the history of Midgard may find the Maemo-Midgard feasibility study worth a read.

For us this was a great fit: many Midgard developers were already fans of Maemo, and Nokia's plan was to run the community infrastructure in an open manner. This meant most of the code was made Open Source, and when possible contributed to Midgard and other upstream projects.

We were joined by excellent people like Niels, Ferenc, and Reggie, and have been running the community infrastructure ever since. In time we would introduce useful new services like social news aggregation, karma, brainstorming, and the hugely succesful Maemo Downloads of which I will talk more later.

Maemo becomes more mainstream

While the 770 had clearly been a hacker device, it was soon followed by the N800 with a cool 70s retro design. While I don't know the actual sales numbers, the N800 was advertised quite actively, and held the sixth place in Amazon sales rankings for all computers.

This was the tablet that really got creativity flowing. We quickly brought location-awareness to it with the Plazes social network, and others in the community made blogging on the N800 possible.

DeviceScape brought another feature: WiFi autologins to pass those annoying captive portals many public access points have. While iOS devices now automatically pop up the authentication dialog in such situations, it is weird that even now this isn't a common feature in mobile OSs.

Later on I gave my old N800 to my mother and she wrote about her experiences with it:

Our Internet tablet is present almost imperceptibly in our life from dawn till dusk. It wakes us up in the morning, and tells news in the evening. It is small and stylish, and it mixes well with the environment.

Its utilities include e.g. a browser for surfing the net, Skype for phone calls, and several radios. Its smallness allows it to be enjoyed together, not separately as often happens with our personal computers. Several times a day we look at weather information, and TV programme guides. Also we use it in looking for dogs in need of home, this being the most important project at the moment.

Sounds quite iPad-esque, doesn't it?

Since Internet Tablets were still a new thing, lots of experiments were happening with them. We ported the CouchDB NoSQL database to the device, and it wasn't uncommon to see robots driven by the device:

Universal communicator

The Nokia N810 entered the picture in late 2007, bringing a hardware keyboard to the Maemo land. This would be the device I dubbed the Universal Communicator and would use as my primary computer on travels:

And I'm not talking about vodka this time, but instead about the latest internet tablet from Nokia, the N810. I've now had the device for some weeks, and it has really started to replace the laptop in many situations.

The idea of an universal communicator is a mobile device that can be used to connect with various communication networks including telephone, instant messaging and social networks. After the latest Internet Communications Software Update, the N810 fits the description quite well

I wasn't the only one seeing the future potential of Internet Tablets. Around this time companies like Rhapsody started experimenting with mobile strategies using the Maemo tablets as testbeds.

Mozilla was also finding their mobile story with their Fennec browser on the N810, a development that would later on result in their highly promising Firefox OS effort. Again we Maemo users were getting a preview of the future.

Even with all this, the communication features of the N810 were a key. I still fondly remember having a N810-to-N810 video call from an Internet cafe in Amsterdam with my girlfriend after my phone had been stolen in the Gran Canaria Desktop Summit...

GeoClue and Summer of Code

Another big feature in N810 was that it had a GPS. Suddenly we would have Nokia Maps everywhere.

This was the time when I was very interested in location-based services. By knowing where it was, a device could act as a travel guide, help to sort out logistics, guide people through public transport, and much more.

Soon there appeared an opportunity to push these ideas forward:

I was a mentor in the maemo project of Google SoC 2007. The project I mentored was Jussi Kukkonen's work on porting GeoClue to Maemo. While in the end Nokia decided to go proprietary with their positioning framework, as result of the SoC project we got GeoClue into GNOME Mobile and Jussi got a job in the field. Via the GNOME Mobile stack, GeoClue is already in at least one device on the market

As the interest in GeoClue grew, I would spend the next couple of years traveling to conferences to give talks on it. Eventually GeoClue would be adopted by most major Linux distributions, and MeeGo.

While some of the more ambitious ideas would only emerge in later services like Google Now, location awareness enabled our tablets to do interesting things, like mobile map making.

Step 4 of 5

2009 was a big shift in Maemo land through the release of the Nokia N900. Now our beloved tablet had become a smartphone! And not just any kind of smartphone, but a fully open and hackable one:

Maemo 5 is a computer platform that happens to fit in your pocket: OMAP Linux Kernel, Xorg server, GStreamer, Telepathy, Tracker, GTK+ (Qt also available) and many more. The telephony stack is also there and SMS also works, but this doesn’t mean that Maemo is now transformed into a smartphone platform. Landscape mode by default, 800×480 amazing display, full qwerty hardware keyboard, a Mozilla based browser providing you the WWW as you are used to get it… We are just expanding the concept of what features a computer that is always with you is supposed to have.

If freedom is your concern then you don’t need to “unlock” or “jailbreak” Maemo 5. From installing an application to getting root access, it’s you who decide. We trust you, and at the end it’s your device. Nokia also trusts the open source community in general and the Maemo community particularly helping in getting casual users through the experience path. The N900 might just be a new and successful entry point for a new wave of open source users and developers.

While the N900 brought a rift into the Maemo community through the older Internet Tablets not being upgradeable to the latest software, it was also an amazing device. With it you would a always-connected device in your pocket with a desktop-grade browser and a full Linux system on it.

For the next couple of years, this would be the device I would carry everywhere, and do almost everything with. I had a full offline copy of Wikipedia to guide me, it talked to Qaiku, the social network we were working on back then.

And most importantly, it had a great camera. With the Frankencamera drivers the phone would often take much better pictures than my real camera. Or what would you say of the depth in this one?

Despite all the advances in smartphone hardware and software since 2009, I still know many people who couldn't consider using any other phone than the N900. This was and is the only smartphone that you could actually make your own.

Maemo Downloads

The N900 was the time when Maemo Downloads really got the chance to shine. Here we had an excellent smartphone, but with a rather lacking official application offering. And so the community stepped in, and started producing a wide range of interesting apps for the device.

One innovative aspect of the Downloads service was how we did quality assurance:

We had a session about application QA in last weekend's GSoC Mentor Summit. I explained how the Maemo Downloads approval process works in a completely open, crowdsourced way. This differs from many distributions where approval of new packages involves obscure decisions and secret handshakes.

The service became quite popular, and by February 2010 we were able to celebrate serving 3.5 million downloads. And this was by far not the end. As of today, the total download count stands at 124 million, with the most popular apps standing around 1 million. Not bad for one device!

Since then, the Maemo Downloads service was also ported for MeeGo with a slick on-device installer client powered by Open Collaboration Services.

Enter MeeGo

In February 2010 I was returning on an early flight from London. When we landed to Helsinki, an ominous SMS was waiting for me on the phone:

Get online. This will wake you up

While I had been traveling, Nokia and Intel had announced joining forces to form MeeGo, a new mobile Operating System. This was a time when nobody was sure what would happen. Would we shut down Maemo and migrate everybody to the MeeGo infrastructure? What infrastructure would MeeGo need? Would there be a common user identity?

Eventually much of these questions would get sorted out. MeeGo would largely run its own infrastructure, but with several of the good parts, like Social News and Downloads copied over.

One effect of the MeeGo transition was that conferences became more grandiose. Suddenly we were holding presentations in football stadiums and major hotels, with the conference parties having whole breweries booked for them. And conference handouts were getting better as well, with people getting tablets and MeeGo netbooks to develop with. The going joke was that next year we'll all get a MeeGo-powered car.

At the same time, MeeGo's wider focus meant that suddenly we were talking about adapting our software to very different environments, from big tablets to smartphones and in-vehicle infotainment systems. We even deployed MeeGo on interactive information displays:

A major part of MeeGo was Nokia's ongoing Harmattan work, which would eventually produce the N9, a widely-praised and slick all-touch smartphone:

The N9 is flawed and doomed, but you have to understand, I don’t care. The overriding experience of using this phone is one of delight and desire. Yes, it can get bamboozled and freeze up, and no, you won’t be finding an avalanche of awesome new apps for it, but those downsides fade in comparison to the abundance of positives. The Harmattan UI is fresh, slick, and as natural as anything the smartphone world has yet introduced, while the physical design is unmatched. Not even the shiny new iPhone 4S feels as luxurious in the hand as the N9. I started off by comparing Nokia’s latest handset to a supercar and the parallels run deep. Like Italy’s finest mechanical produce, the N9 won’t be found in many shops, has a tendency to break down, and inspires an emotional rather than pragmatical response. There’s an added underdog charm in knowing it has been discarded by its maker and deemed unworthy to carry the Nokia crown.

For much of its time, MeeGo was widely misunderstood as Android without Java. I sought to set the record straight in my Understanding MeeGo post from June 2011:

MeeGo is much more than just handsets or tablets. It is an attempt at creating a standardized industrial Linux distribution that can be used anywhere from in-vehicle infotainment devices to TVs to, indeed, handsets.

It is a true open and collaborative environment, managed by Linux Foundation. The governance model is there to ensure that MeeGo stays a vendor-neutral platform that anybody can build their products on top.

Empire Strikes Back, the mobile Linux winter

By the late summer of 2011 both MeeGo and webOS were dead. MeeGo because Nokia entered the world of Redmond, and webOS largely because HP couldn't decide what they wanted to be when they grew up.

Tizen was launched on the ashes of MeeGo, with essentially similar plans and ambitions, but with every instance of Nokia and Qt replaced with Samsung and EFL, and obviously causing yet another costly rewrite. I can only imagine how industrial vendors like car manufacturers felt, with long-advanced MeeGo based product plans and the rug suddenly pulled from under them. However, the jury is still out on whether Tizen will succeed or not.

Ever since the open mobile ecosystems of were killed, I've been appalled of the direction this "Post-PC era" is taking us. Through patent wars and locked-down app stores, the world of mobile software is becoming a power play where some win big and everybody else loses.

This is largely the reason why I now carry a humble Symbian device instead of the current breed of smartphones. The existing mobile ecosystems are either too dystopic or in turmoil for me to get involved in any of them.

App Stores are the new record labels

As much of software is about disintermediation, of making the world run more smoothly through removal of middlemen, it is interesting that we software developers are now driving ourselves to a world full of middlemen. A world where we suddenly have to ask for a permission to do something new.

In a world where everything must go through the rules and regulations of an app store without any oversight we, the developers, will suddenly be in the same abused stage as artists are with their labels. We take all the risk and all the effort on building software for our users. The middleman then can invalidate all our hard work by arbitrarily making it impossible for their ecosystem to run the app. And even if they do accept the software, they'll take a hefty cut of the proceeds.

How can this make sense to an independent developer?

Universal runtimes

Many of us have started our programming careers in a free world. Whether we were building software for Linux, Windows, Mac, or the Web, we could largely do anything there. We could publish the software under a free license, charge license fees, or provide it as a rental service. And the only limiting factor would be whether our potential users liked the software or not.

The fever of a mobile app gold rush has made many throw away those freedoms. With 60% of developers losing money in the process this seems hardly worth it.

As desktop and mobile are being locked up behind anti-competitive restrictions, often justified in the name of security, many developers are drawn to the last open frontier, the Web. I've addressed this in more detail in The Universal Runtime, but Y-Combinator's Paul Graham said it well already back in 2001:

How will it all play out? I don't know. And you don't have to know if you bet on Web-based applications. No one can break that without breaking browsing. The Web may not be the only way to deliver software, but it's one that works now and will continue to work for a long time. Web-based applications are cheap to develop, and easy for even the smallest startup to deliver. They're a lot of work, and of a particularly stressful kind, but that only makes the odds better for startups.

Return of the MeeGo

For supporters of free software, the last year has been looking quite grim. Desktop becoming irrelevant, mobile becoming more and more closed. Instead of quietly giving up, some saw this as an opportunity to fight back and build something new. In Where is the future for openness in mobile, I wrote:

The other is to take the matters in our own hands. There is precedent for this. Much of early Linux activity came from the efforts of the community, not on the initiative of corporate interests. And there have been OpenMoko and Mer, the latter an attempt to make a fully open version of Nokia's Maemo environment, suspended when MeeGo promised to bring the same benefits.

Well, now Mer is back.

In FOSDEM last winter I helped to organize a Open Mobile Linux track with the goal of bringing together the people involved in building an open version of the Post-PC future. I thought that with what had happened to MeeGo and webOS, there would be very little interest. Instead, the track was very lively, and we even had to turn people away to remain within the limits of fire safety regulations!

While the common view seems to be that the future is shared between Android and iOS, there have been some bold attempts at tablets powered by free software. And of course there is today's big announcement:

Jolla

I had been hearing rumors of something happening in the post-MeeGo space for a while, and today it seems it was finally time to make things public. The JollaMobile Twitter account went live with the following announcement:

As can be expected, this brought a lot of notice on Twitter, as well as on Maemo forums and tech blogs. Much of the Jolla plans are still secret, and there isn't the kind of community infrastructure we from Maemo have been used to. But I'm sure these areas will become more clear as time progresses.

From Jolla's first press release:

Nokia created something wonderful – the world's best smartphone product. It deserves to be continued, and we will do that together with all the bright and gifted people contributing to the MeeGo success story.

Together with international investors and partners, Jolla Ltd. will design, develop and sell new MeeGo based smartphones. The Jolla team consists of a substantial number of MeeGo’s core engineers and directors, and is aggressively hiring the top MeeGo talent to contribute to the next generation smartphone production.

Indeed it appears Jolla has learned many lessons from the earlier stages of the MeeGo story. Instead of reinventing all the wheels, they're working together with many of the existing players, including the Mer Project. Carsten Munk explains:

And for the record, JollaMobile is not Mer. Mer is not JollaMobile. JollaMobile uses and contributes to Mer. Mer is a mobile core (without UIs and Hardware adaptations), usable and used by many different companies, with many different contributors.

Will all of this work? Who knows, the world of mobile is a difficult one, with many failed efforts in its history. The things going for Jolla are a the excellent technical base of Linux, Mer, and Qt, the battle-scarred people who were able to produce the excellent Nokia N9 phone, and hopefully the existing Maemo community.

I wish them luck in their endeavors, and look forward to the day when I can again proudly show my MeeGon:

I've been writing this story on a stormy Berlin weekend, wearing my Mer Project T-shirt, and watching my Twitter feed fill up with supportive comments on Jolla and Mer. Exciting times!

Kinect Air Cursor: Let your hand be the mouse

henri.bergius@iki.fi (Henri Bergius) — Wed, 27 Jun 2012 00:00:00 -0700

If today's Google I/O keynote where they parachuted to the conference center from a Zeppelin while streaming the whole experience on a Hangout via Project Glass wasn't enough future for you, here is another thing.

As part of the SmarcoS project, we've been working on making the Kinect work as an input device for Qt applications. Basically you move your hand in the air, and are able to grab and drop things on the screen.

We call this the Air Cursor. Here is a quick video of manipulating a simple HTML5 application with it:

Now, this may not be the way you want to control the computer you're working with the whole day. Instead, we see this sort of interface as very useful for large displays in meeting rooms and public spaces.

Instead of a touchscreen that easily gets messy and requires people to stand in front of it, with the air cursor you can use a regular TV or projector, and use your hands to manipulate the information on it. The gestures we use are natural enough that everybody we've had trying the tool has figured them out in matter of seconds.

Our Qt Air Cursor is free software under the LGPL license, and is built on top of the OpenNI library, with OpenCV used for recognizing the grab gestures.

I believe this is a great start for using natural interaction to control information software or multimedia applications. Simple gestures like grab-and-drop and swipes work, but there is still a lot of UX territory left to explore.

If you have ideas where this sort of new input techniques could be used, feel free to get in touch. Or simply to integrate the Qt Air Cursor library into your applications.

The Qt Air Cursor was demoed for the first time in this year's Qt Contributor Summit in Berlin. Our simple "Grab to the Future" example game gathered quite a large audience, with the high score ending up at a respectable 18. You know you're doing something right when the event catering staff also wants to try your input device demo.

How do I turn this thing off?

henri.bergius@iki.fi (Henri Bergius) — Fri, 01 Jun 2012 00:00:00 -0700

Michael Mace has a very interesting review of Windows 8. As whole it is a good read for anybody interested in where this whole modern computing thing is going. But this part specifically caught my interest:

How do I turn this thing off? The concept of a power button is pretty central to any electronic device. You turn it on in order to use it, and you turn it off when you're done. It's easy to turn on a Windows 8 computer; you just press the power button on the computer. But pressing the button again does not turn off the computer. Instead, it puts the computer to sleep.

Sleep is a good thing in a computer. It lets a computer restart quickly, and keeps your apps active. But it does consume power, which is an issue for ecologically-conscious desktop users, and a primary concern for laptop users. Also, I find that it's helpful to turn off Windows from time to time because the OS gradually becomes confused and slow as you launch and quit large numbers of apps. So I expect to be able to easily turn off my computer, and I think most Windows users will feel the same way.

It is absurdly difficult to turn off Windows 8. So difficult that there are entire web pages devoted to tutorials on how to do it

This discussion should be familiar to users of GNOME3.

Personally I don't find this shutdown discussion very interesting. I maybe shut down my laptop once a year, when going on a holiday trip that the battery wouldn't last through. And as there is a whole hardware button dedicated to shutting down the computer, why clutter the UI with one?

In general, it seems that Microsoft is headed for their own GNOME3/KDE4 moment, where the desktop resembles the old thing while trying to introduce new paradigms. These sort of disruptions can be risky, as Michael notes:

The rollout of Windows 8 has very important implications for not just Microsoft but everyone in the tech industry. In normal times, most people are unwilling to reconsider the basic decisions they have made about operating system and applications. They've spent a huge amount of time learning how to use the system, and the last thing they want to do is start learning all over again. That's why the market share of a standard like Windows is so stable over time. But when a platform makes a major transition, people are forced to stop and reconsider their purchase. They're going to have to learn something new anyway, so for a brief moment they are open to possibly switching to something else. The more relearning people have to do, the more willing they are to switch. Rapid changes in OS and app market share usually happen during transitions like this.

...

Although Microsoft calls it Windows, and a lot of Windows code may still be present under the hood, Windows 8 is a completely new operating system in every way that matters to users. It looks different, it works differently, and it forces you to re-learn much of what you know today about computers. From a user perspective, Microsoft Windows is being killed this fall and replaced by an entirely new OS that has a Windows 7 emulator tacked onto it.

This kind of shift could either mean a great opportunity for alternative computing environments like the various Linux desktops and OS X, or it could move people much faster towards the post-PC paradigm of simpler computing devices connected to cloud services. For the latter the free world has little to offer.

Open Advice

henri.bergius@iki.fi (Henri Bergius) — Mon, 19 Mar 2012 00:00:00 -0700

I seem to have not blogged about this, but Open Advice, our book on Free and Open Source Software: what we wish we had known when we started, was published last month.

The book was edited by Lydia Pintscher and includes essays from 42 authors, many of whom you'll recognize if you tend to go to FOSS conferences. The LWN book review concludes:

Open Advice is a book that will be helpful to those who are new to FOSS, but, because of the individual voices, styles, and tones, it doesn't read like a "how to". It could even be recommended to those who aren't necessarily interested in contributing, but are curious about what this "free software thing" is all about. It is, in short, a great book for a variety of audiences and the (mostly) two or three page essays make it easy to read, while the anecdotes and recollections personalize it. The authors, editor, and everyone else who helped should be very pleased with the result. Readers will be too.

I probably shouldn't give the ending away, but my essay on cross-project collaboration, a subject I've also blogged about, ends with:

Good luck with breaking down the project boundaries! In most cases it works if your ideas are good and presented with an open mind. But even if you do not find a common ground, as long as your implementation solves the use case for you it has not been in vain. After all, delivering software, and delivering great user experience is what counts.

The book is licensed under CC-BY-SA, and is available as free download in ePub, mobi and PDF formats, and as paperback from Lulu. The book sources are available on GitHub, patches welcome!