W3C home > Mailing lists > Public > public-css-testsuite@w3.org > March 2009

RE: attribute-value-selector-004.xht not well formed

From: Sylvain Galineau <sylvaing@microsoft.com>
Date: Wed, 11 Mar 2009 07:50:41 -0700
To: Anne van Kesteren <annevk@opera.com>, "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>
Message-ID: <5D97C7EB4695104AB6345E56FE356B19408A7B44C9@NA-EXMSG-C125.redmond.corp.microsoft.com>
> Because not matching [1badattr] is dealt with at the error handling level  
> and nowhere else.

Error handling is not what is tested here. These tests do not assume your error handling is correct.
> That [1badattr] does not select anything follows from parsing, not  
> selection. I.e. you would be able to select a 1badattr attribute if you  
> use CSS escapes.

It follows from parsing if the implementation correctly implements all of it, including error handling, which are not targeted by this test. 

> I don't see it. 

Well, that we can agree on :)

> You're still testing CSS parsing as far as I can tell, 

Granted, any CSS testcase verifies parsing to some extent. These do not verify on parsing rules though.

> If this was really about selection  
> tests only I think you should not include rules that are thrown out due to  
> syntax errors. Those can be moved to the syntax section.

Why shouldn't selection tests validate the cases that should fail as well as those that should pass ? It does not matter why a selector should work or not in this part of the test suite. Your reasoning would be totally fine with me if the tests in this chapter assumed the implementation handles error handling properly. They do not. They do not even need to care *why* a selector shouldn't match. Only that it doesn't.

Received on Wednesday, 11 March 2009 14:51:59 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:13:18 UTC