You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Create more machine readable source of truth for mapping tables - use for tests also #258
The more I think about this, the more I think it won't be easy to automate the writing of tests from the content in the Core-AAM tables alone. For example, for each table, for the tests you would also need to provide the html markup to test. Also there is a fair amount of prose in the Core-AAM tables, and maybe you could try to write that a bit more programmatically... but I'm not altogether sure this is worth the effort. In a way, it's more like the table can be generated from the tests than the other way around, and there will still be (probably) some untestable statements, or "authors SHOULD" statements.
The AAM tables are not modified often, when they are modified, it won't be a lot of work to update the tests. So maybe automating the updating of the tests is not worth the effort need to create the right abstraction. According to the helpful xkcd chart I can spend 1-2 days if in the future it saves 30mins to 1 hour of work writing tests a month.
Activity
spectranaut commentedon Mar 3, 2026
The more I think about this, the more I think it won't be easy to automate the writing of tests from the content in the Core-AAM tables alone. For example, for each table, for the tests you would also need to provide the html markup to test. Also there is a fair amount of prose in the Core-AAM tables, and maybe you could try to write that a bit more programmatically... but I'm not altogether sure this is worth the effort. In a way, it's more like the table can be generated from the tests than the other way around, and there will still be (probably) some untestable statements, or "authors SHOULD" statements.
The AAM tables are not modified often, when they are modified, it won't be a lot of work to update the tests. So maybe automating the updating of the tests is not worth the effort need to create the right abstraction. According to the helpful xkcd chart I can spend 1-2 days if in the future it saves 30mins to 1 hour of work writing tests a month.
scottaohara commentedon Mar 5, 2026
though maybe not the intention of this issue, there is some overlap to this idea that would help with w3c/html-aam#122
the challenge also being that some tables aren't 1:1 with core aam, as there might be additional mapping details or comments for an HTML element.
pkra commentedon Mar 11, 2026
Discussed in today's editors call https://www.w3.org/2026/03/11-aria-editors-minutes.html#72ef