Documentation

Documentation for developers

Documentation for users who want to fiddle with the alphabets, colours, or training texts

training text

We strongly encourage users to optimize the training text for their personal use. Dasher learns all the time as you write, and saves what it learns in a file (for example, all the writing of a linux user using English will be stored in a file called ~/.dasher/training_english_GB.txt). It is a great idea to take a load of text that is like what you are going to write in the future and put it into this file, or into the equivalent system-wide file (for example, /usr/local/share/dasher/training_english_GB.txt on a linux system). Next time Dasher is started, or next time you switch alphabet, both these files will be read so as to train the language model to make appropriate predictions.

colours

Users can configure the colours used by Dasher. Dasher uses 242 colours at any one time. How those colours are defined and used is specified by two files, the alphabet file (Example) and the colour file (example). Both these XML files are plain text files that you can edit. The alphabet file specifies by numbers between 1 and 242 which colours each letter and group in the alphabet should have. The colour file defines what those 242 colours actually are.

Version 4 of Dasher offers several colour schemes to choose among. Many of our alphabets use a colour scheme (a 'palette') called European/Asian. Here's how the default colours work for European languages. The GROUPS are coloured like this:

• accents and marks - ORANGE (colour number 51)
• lower case characters - this group is not coloured
• upper case characters - YELLOW (colour number 111)
• numerals - RED (colour 113)
• punctuation - DARK GREEN (colour 112)
• space character and newline - WHITE (not actually a group)