Gecko exposes nodes styled as a table (using CSS "display: table") as tables to accessibility APIs. Also, Gecko does not expose the IAccessibleText interface on tables. This makes text inaccessible (except via text leaf nodes) when there are text children of a CSS table. Str: 1. Open the attached test case. 2. Try to retrieve the text of the table accessible. Actual: No text interface is exposed, so no text can be retrieved. Expected: There are three possibilities: 1. The text interface should be exposed and the text content ("table cell") should be exposed; 2. The text content should be exposed on another node beneath the table, perhaps with a role of text frame; or 3. The node shouldn't be exposed as a table to accessibility APIs at all. This could probably be considered an authoring error: it doesn't make sense to use "display: table" without using "display: table-cell" on children. However, we've now seen two separate cases of this in the wild. The result is that NVDA doesn't render the text children at all. See these related NVDA tickets: http://www.nvda-project.org/ticket/1373 http://www.nvda-project.org/ticket/1338 This occurs in both Firefox 3.6 and 4pre.

taking into account http://www.w3.org/TR/CSS2/tables.html#anonymous-boxes we should follow 2nd I think and therefore it should be a dupe of bug 576838 (marking a11yfx5 blocking to make sure they are fixed both).

Ping? It turns out that this bug appears here: http://openelec.tv/forum/90-miscellaneous/73147-openelec-for-amlogic-s805-quad-core-stb?start=105 In that instance, the author does use "display: table-cell" for the forum post author. But the content of the post lacks that, but it is contained in an element with "display: table". As a result, there is a table which has either no children, or children just for the links. The text itself is not accessible.

Hello, With Orca I can see that string. So has this bug an important user impact? Regards