This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
When changing the test-suite metadata you only have choices of html, svg and xhtml. I have more formats, which sometimes require their own tests (eg. character encoding can differ for slightly across all formats) and sometimes have different results, and therefore need to be distinguishable in the results listings. I've played around with naming conventions a bit for the i18n test suite. Could I suggest we reserve the following names: For HTML4 html4 For HTML5 html5 For XHTML 1.0 served as text/html xhtml For XHTML 1.0 served as app/xhtml+xml xhtmlx For XHTML5 xhtml5 For XHTML1.1 xhtml11 (the existing html and xhtml could be used for two of these formats, if needed)
Why? I don't see how these extra formats add anything.
(In reply to comment #1) > Why? I don't see how these extra formats add anything. See https://www.w3.org/Bugs/Public/show_bug.cgi?id=15062#c8
Why do you want to test what HTML4 requires? I sure hope nobody is trying to implement it.
The tests we create are aimed at discovering how well i18n features of a particular format are supported by user agents. The answers supplement our outreach efforts, and they have lead to browser developers improving their support.
(In reply to comment #4) > The tests we create are aimed at discovering how well i18n features of a > particular format are supported by user agents. The answers supplement our > outreach efforts, and they have lead to browser developers improving their > support. I don't think that answers the question. Are browsers implementing i18n features present in HTML4 but not HTML5? If they are, and these features are important enough to test, they be pulled into HTML5.
We try to be explicit in our assertions as to whether a test is exploratory or not. Mostly, the difference is dependent on whether the test is served as html or xml, but there are a few special cases. So, for example, to check whether <meta charset='utf-8'> works in html4 as well as in html5 we have separate tests (since it's not specified by html4 and the html4 test can therefore only be seen as exploratory). A similar distinction is made, in the other direction, when testing for the effect of the charset attribute (which is not part of HTML5). You can see examples of how we map tests to formats at http://www.w3.org/International/tests/html-css/list-encoding and http://www.w3.org/International/tests/html-css/list-language (Although those pages allow you to run a test in a number of formats, we probably only want to add a few special cases to the w3c test harness.) Another example would be ruby markup, which is supposed to be supported for XHTML 1.1 only prior to HTML5, and complex ruby not even in html5. See http://www.w3.org/International/tests/html-css/list-ruby
(In reply to comment #6) > So, for example, to check whether <meta charset='utf-8'> works in html4 as well > as in html5 we have separate tests (since it's not specified by html4 But HTML5 does specify that it needs to work in documents with a HTML4 doctype.