Possibly the most useful is being able to serve different content based on network bandwidth and latency. If on a limited data plan, you could tell websites you don’t want want HD content by default. Mobile users with high bandwidth but a poor network latency may have content adjusted for them in a different way to someone connected via a low-latency, low-bandwidth ethernet link.

This would allow device pixel media queries to be used just for screen sizes, without making assumptions that would end up showing a full site to someone tethering their laptop to a GPRS connection or a cut-down version to someone who happened to resize their browser.

In addition to network conditions (and even preferences), a whole host of other capabilities could be determined via media queries. A hover menu could be limited to users with a mouse. Keyboard shortcuts could be shown to users with hardware keyboards. Complex animations could be disabled by default for users with lower-spec CPUs or GPUs.

This finer grained view of the user would reduce the assumptions currently made by lots of websites and allow website authors to provide sensible defaults to users without having to assume context.

In terms of getting these implemented, the next steps would probably be to have discussion on W3C/WHATWG mailing lists and file some bugs in browsers. Would anyone like to help take responsive design further?

The code is on GitHub and it’s under a Creative Commons Attribution 3.0 License. It’s tested on Android and iOS but if you have any questions or issues on jQuery Fast Click, please use GitHub’s issues.

Traits are a new feature of PHP 5.4 (released in 2012), which allow horizontal reuse of code in class. This is similar to multiple inheritance (eg as in Java) but more similar in implementation to mixins in Ruby.

TaskThere lets you quickly assign tasks to your favourite GPS locations. Add “buy cheese” at “Supermarket” or “mow lawn” to “Garden”. When you add a new place, TaskThere looks up your geolocation from either your mobile’s GPS, nearby cell phone towers or WiFi hotspots. As you add more places, TaskThere will automatically switch to your nearest one to show you what tasks you have near. You can manually switch locations or disable the GPS if you want to save battery life.

TaskThere is a web app, built using HTML5 JavaScript APIs such as Geolocation and Local Storage. It runs on any modern web browser, including mobile browsers. All your data is stored locally so there’s no privacy issues. TaskThere is lightweight so it works fine on 2G and once you’ve loaded the app your network connection isn’t required any more.

All the code to TaskThere is on GitHub. It works great but it’s a bit rough round the edges. It needs some design work and there’s a few features I’d like to implement. If you’d like to get involved, contact me on GitHub or just submit a pull request.

Please let me know Iif you’ve used TasktThere and have any feedback.

More thoughts and comments about this over on the Hacker News discussion.

My first memory of Linux was when I was about 12. My 486 wasn’t fast enough to play MP3s in Windows. Previously, if I wanted to listen to a song, I would convert it to a WAV on our family Pentium 75, split it into 1.4MB chunks and copy it via ~15 floppy disks onto my PC.
I installed RedHat 5 (from a PC mag CD). RedHat came with a commandline MP3 player (called mpg123) that would decode and play MP3s on my 486. This meant it only took 2 or 3 floppy disks to copy a MP3 I’d downloaded from our family PC’s 14.4kb/s net connection.
Today, my phone runs Linux (Maemo), my work PC runs Linux (Ubuntu), my laptop & our TV run Linux (Ubuntu), our router & NAS run Linux (some kind of Debian). Our next car is likely to run Linux and so might our fridge. Thanks Linus!

Annotations are semi-structured as there is no formal registration process for the namespace and key. Sitepoint have a good article on the Twitter annotation format. This post is an attempt to collate some annotation ideas into a unified place.

There are already some annotations contained in tweets. These include:

• username
• in-reply-to tweet
• timestamp
• location
• contributors

Here are some more. Some day I may clean this up and try to define a standard format and namespace.

• Mood (happy, sad)
• Changed mood (before-after)
• Weather
• Privacy level
• Image/video/sound clip URL
• General URL
• Licence (how can other people sue your tweet?)
• Copyright holder
• Distribution (global / specific country)
• Joke
• Source / attribution
• Current transport (walking, driving, train, etc)
• Traffic
• Split personality
• Translation (including to non-geek)
• Product price
• Special offer / discount
• For sale / wanted
• Currency
• Rating
• Event date/time
• Topic or category (if you generally tweet about a few categorizable topics)
• Group ID (many tweets in the same group could be shown as 1 in supporting clients)
• Vote / poll ID (vote for an option in a poll)
• Proxy (allow other people to use your account, like cotags)
• Question / answer
• Listening to artist / album / song

Feel free to add your own ideas in the comments.