Edbrowse and accessibility on the Web

One approach at making the Web accessible to people with vision impairments is to introduce support for screen readers, braille displays and keyboard shortcuts into pre-existing Web user agents, which generally employ point-and-click interfaces. The advantage is clear: hundreds of millions of lines of complex software, in the form of browsers, mail user agents, etc., can be ‘reused’ simply by adapting the screen. Some have argued this is the only practical approach.

An alternative, which we find mostly overlooked, is to develop such agents, and indeed software useful for day-to-day tasks in general, around user interfaces that are already convenient for a person with vision impairments.

We believe that the traditional command-line interface is the most practical interface in this regard. This is especially relevant for the deaf-blind community, who use a small braille display. Screen readers, which have attained some level of success at bridging the gap, are unsuitable in this case. The braille interface is intrinsically linear, i. e. a line of mechanical braille cells, thus the adapter must somehow squash a two-dimensional browser, be it GNU IceCat or even Lynx, into a command-line interface. Other blind users, who have access to screen readers, still find the ‘adapted’ experience suboptimal.

A new command-line application, designed from the ground up with minimal input and output in mind, is more appropriate for these linear interfaces.

Edbrowse provides an example of a mature Web user agent developed over the past 20 years along the ideas outlined above. Its highlights include:

• command-line interface, modelled after the traditional Ed editor and grown to include advanced features, such as multiple buffers and user-defined functions;
• Javascript support, based on the Duktape engine;
• integrated email (SMTP, IMAP, POP3) and database (ODBC) clients;
• integrated directory editor;
• fully functional as an editor proper.

Nevertheless, support for modern Web technology, and first and foremost for the latest advances in Javascript, is largely a moving target, and likely requires more effort than the current team of developers can provide. As such, and given the importance we see in its aims, we believe that Edbrowse should be considered for inclusion in the forthcoming revision of the list of high priority free software projects.