Dasher – Dasher Blog

Blog: Developer Update June

Mon, 15 Jun 2020 00:00:00 +0000

Hello Dasher Fanatics!

This is the first in a series of developer updates that will be published a few times a month to keep those interested informed of Dasher progress. I will mostly attempt to keep the information targeted and concise.

Repository Restructure nearly Complete

A special shout out to Jim Hawkins, who has implemented the pure Javascript version of Dasher that all of this work is now based upon. In order to keep the pure JS version clean, Jim took the time to abstract out platform specific components of the application. The new structure consists of:

Repository Rename(s) Incoming!

In order to comply with trademark issues, we will be changing the name of the https://github.com/dasher-project/redash repository to https://github.com/dasher-project/dasher-web In addition, we will be renaming the “master” branch to “main”. If your project depends on this repository and you are worried about the timing of this change, please let us know sooner rather than later via the contact information below.

Mozolm Proof of Concept

The Mozolm integration proof of concept has been completed and marks a big move forward in allowing for more complex language models to be utilized in a variety of scenarios.

NLNet Milestone Series 1 Complete

Developers completed work on the first series of features suppored by NLNet. This has included migrating documentation to the Github Wiki, drafting a user experience survey, and creating a Dasher desktop proof of concept. If you are a user or trainer of Dasher, please help us make it better by taking the appropriate survey linked below!

Thanks for reading!

Interested in supporting us? We’d love to find a way for you to contribute! If there is anything I missed, feel free to message me directly and I can amend this newsletter or include it in the next one. Adam Spickard (adam@technoke.com)

Blog: A new website is here

Wed, 01 Jan 2020 00:00:00 +0000

So a group of us have been working on Dasher v6 for a little while now - and we figured its about time to get something out there publicly about the inner workings of the project - so you can help out or just find out where this is all up to.

The site has a little history. Back in early 2019 I was trying to update the documentation for a number of end users and we attempted to move all the documentation from the inference groups site (there is a lot of it) - to a new markdown format. We did it but I then shelved the project - as moving the docs was one thing but they all needed a little more sprucing up.

Skip forward some time and the development group were feeling that it was time for a public presence to share our news and updates - but also get other peoples input. We decided to strip out the docs for now and focus this site purely on the new version 6 build.

The site is using docsy - a fab Google product using hugo as a static site generator for documentation. There are other Static site generators (e.g. VuePress) but for this task Docsy seemed like a good fit.

Now I’m no UX expert. If you have any suggestions let me know and I’ll do what I can!

Hopefully the team can input there own thoughts and discussions when they get the chance. Otherwise I’ll try and keep it all updated as we can.

Screenshot of first build of site
Photo: Will Wade / CC-BY-CA

Blog: Announcing Dasher-v6

Wed, 01 Jan 2020 00:00:00 +0000

Steve Saling uses Dasher with a headmouse.
Photo: Jesse Costa/WBUR

History

Dasher started life in the form of a prototype written in 1997 by David MacKay, who at the time was a lecturer in physics at the University of Cambridge. David had been discussing the then current generation of mobile devices with a colleague, and felt that the use of physical keyboards for text entry made them unnecessarily bulky. Dasher was inspired by ‘arithmetic coding’, an efficient file compression technique which was one of the subjects taught by David in his popular fourth year course for undergraduates.

Between 1998 and 2002, one of David MacKay’s PhD students, David Ward, developed Dasher into a fully featured open source application for Windows and Linux as part of his work towards his thesis. Over the years, many other people worked on Dasher, many of whom were in David’s research group, the Inference Group. The software was ported to a number of platforms, including Apple macOS and iOS as well as few which are now less relevant, such as Solaris and various pre-iPhone mobile devices. Dasher became part of the GNOME Desktop project in 2002, which resulted in the software being included in several Linux distributions.

Dasher was conceived as a flexible text entry system. As well as the core applications in AAC and mobile devices, it was designed to be portable and has been used on a number of more specialised hardware platforms, such as wearable computers. Dasher has also been used for text entry in languages with large alphabets, such as Chinese and Japanese.

Dasher was developed to work as a standalone tool, but was also integrated into assistive text entry frameworks on Linux and Windows, which allowed it to be used to enter text into other applications. Since early versions, Dasher has included speech synthesis software to allow it to function as a communication aid.

As well as conventional pointing devices such as mouses, Dasher has been evaluated with eye trackers, joysticks, and discrete buttons, all in various configurations. A number of research papers have been published comparing the effectiveness and efficiency of these input mechanisms.

The original Dasher project was largely dormant after version 4.11, which was released in March 2010. In 2016 the Dasher project was restarted, and version 5 was developed. In 2019 the current project to create version 6, which is a complete rebuild using a more modern technology base, was kicked off.

So why start all over again?

A number of people have worked on Dasher’s code base and going forward - with more features being added, or language model testing being carried out - or just to keep the application updated across the different operating systems - its not been easy. Its certainly not been the easiest codebase for anyone just to jump on board and help out on.

But also - its 2020 and its time to build an architecture that last another 20 years.

What’s Next

There’s a lot to do. Here is a quick rundown:

Build an awesome language model that has an impact for dasher users - but hopefully - the entire AAC ecosystem. We hope to support as many languages as possible.
Create a modular architecture that allows for input (eg touch, switch, headmouse, eyegaze) and output methods (text in a display bar, speech, secondary display) to be accessed by a API/SDK. But also allow for configuration of the zooming interface by API - and make it accessible to the user themselves
Providing an architecture to extend dasher more easily than before. Want to drive Dasher by speech? By Breath? We want to help future proof dasher and help the development pathway easier.
Develop a user experience for the app that allows new users to feel comfortable with dasher and learn - steadily - how to use it - and adapt to their changing needs.
Iterative user testing throughout creating a solution that is user led.

Getting Started

If you want to get started, have a look at the issue queue - and submit a Pull Request.

What do we need?

If you are a user of dasher - please get in touch with Will. As you can see here - we are trying to pull together as many current use cases to identify priorities - but also help with user testing

The team coding dasher is all working on this pro-bono. A massive, massive thanks to them. However we always need more coding support. If something in the issue queue doesnt look right - then please join the community or drop us a email and we’d love to point you in the right direction to get you helping out. We do need to pay for user testing and design work. If you can offer funds towards the project - please get in touch with Will. If you are a UX/UI expert we particularly want to hear from you!

Blog: Designing a new logo

Wed, 01 Jan 2020 00:00:00 +0000

Colours and Inspiration for a Dasher logo for 2020
Designs: Fig/Ace Centre 2020

Whilst the hard work of language models and toolchains and whatnot are worked on by some very clever people there are some jobs that seem a lot easier - but possibly may cause a lot of contention!

The original dasher looked like this:

Original logo - remade by the Ace Centre for the iOS release.
Designs: Fig/Ace Centre 2020

Here at the Ace Centre - we had to recreate this logo with vectors for the release of the updated iOS build (we didnt do much - largely just got it working)

But going forward we felt we needed to update this to differentiate this completely new build and make it a bit “fresher”.

The above designs were some suggestions. In the end we have plumped for the logo we are using. But do get in touch if you can do better or have any feelings. We’d love to hear from you!