Lessons learned from a project by Hans Hillen and Jennifer Gauvreau, what should have worked that didn’t! Lots of JAWs/IE bugs and how they got around them.
Their slides and demos are up which is great, so I won’t do detailed notes. Building on that I took away a few things:
They went through three versions of JAWs during the project, simply because it was a long project!
Landmarks with JAWs (12) was a bit, um, quirky. After following a main landmark, it would read out all the content of the landmark (the content) in front of each form field. This has been fixed in more recent versions.
A lot of the issues found relate to IE. Firefox (even with Jaws) tends to be much better than IE these days. Unfortunately, not many end-users would know that…
They ran into the problem of pop-over dialogues not allowing screen reader users to read the content when role="dialogue" is applied. However, the most recent version of NVDA solves this by allowing you to switch browse mode with NVDA+space. Their solution was to wrap the content in a role="document".
Someone in the audience mentioned that the WAI-ARIA spec will be updated to make it less confusing, and say that the user-agent doesn’t need to switch to application-mode.
Also, when a pop-over is open, the JAWs links-list allows you to get into the background links and content, which is problematic. They got around that by sectioning it off with aria-hidden.
A lot of the issues were found when real content went into the application, boilerplate text doesn’t really cut it in this situation.
Icon fonts work well in high-contrast mode, but can break when people use their own style sheets to override font-family. It takes quite a lot of work to get around this, scripting fall-backs if you detect high-contrast or images are not loading.
One contribution to “CSUN: Accessibility theory vs implementation reality”
Comments are closed.