5581 2008-03-20 12:04:32 +0000 Several invalid forms of @charset syntax reported as valid in explicit css2.1 grammar - part II 2008-04-03 14:06:45 +0000 1 1 1 Unclassified CSSValidator CSS 2.1 CSS Validator PC Windows XP RESOLVED FIXED P2 normal --- 1 cecil dave.null www-validator-cvs oldest_to_newest 19559 0 cecil 2008-03-20 12:04:32 +0000 The following test cases are mistakenly reported as "valid" against the explicitly selected CSS 2.1 grammar. [ checked at 2008-03-20 against live snapshot of the validator, http://qa-dev.w3.org:8001/css-validator/#validate_by_input+with_options ] -- test case 8: selected grammar=CSS2.1 @charset /**/"utf-8"; Correct response: INVALID; should report INVALID when selected CSS 2.1 grammar Current status: INCORRECT: The validator's response is INCORRECT, as of 2008-03-20. -- -- test case 9: selected grammar=CSS2.1 @charset/**/ "utf-8"; Correct response: INVALID; should report INVALID when selected CSS 2.1 grammar Current status: INCORRECT: The validator's response is INCORRECT, as of 2008-03-20. -- -- test case 10: selected grammar=CSS2.1 @charset "utf-8"/**/; Correct response: INVALID; should report INVALID when selected CSS 2.1 grammar Current status: INCORRECT: The validator's response is INCORRECT, as of 2008-03-20. Regards, Cecil Ward. 19562 1 cecil 2008-03-20 13:05:10 +0000 Note: see discussion ref bug 3381 It now appears that, just as predicted earlier, the parsing of the @charset rule has been adversly affected by changes in respect of the bugfixes for bug 3381. The problem is the current production for the @charset rule in the grammar, which is not 'atomic', not a terminal. It is not ok to allow a lexer to do its usual work in the middle of processing the @charset rule because it is not at liberty to for example collapse multiple items of whitespace nor to eat up comments completely in CSS2.1 mode. The suggested modification to the lexer in respect of @charset described in http://www.w3.org/Style/css2-updates/CR-CSS21-20070719-errata.html is not radical enough. 19637 2 ylafon 2008-03-28 10:14:42 +0000 It would be relatively easy to parse only the strict version of CSS21, however the hard part here is to allow CSS2 constructs ("@charset S* STRING S* ;) while managing to cope with the strict version. 19640 3 ylafon 2008-03-28 15:14:46 +0000 Fixed, all the test cases reporter are rejected as they should. Also there is a special test to have the string starting with " and not ', per http://www.w3.org/TR/CSS21/syndata.html#charset Available on http://qa-dev.w3.org:8001/css-validator/ Thanks, 19652 4 543 ot 2008-03-31 14:55:10 +0000 Created attachment 543 test case 19653 5 544 ot 2008-03-31 14:55:39 +0000 Created attachment 544 test case 2 19654 6 545 ot 2008-03-31 14:56:00 +0000 Created attachment 545 test case 3 19723 7 546 ylafon 2008-04-03 14:06:45 +0000 Created attachment 546 simple quote for charset -> fails 543 2008-03-31 14:55:10 +0000 2008-03-31 14:55:10 +0000 test case charset1.css text/css 48 ot QGNoYXJzZXQgLyoqLyJ1dGYtOCI7CgpwIHsgZm9udC13ZWlnaHQ6IGJvbGRlcjt9 544 2008-03-31 14:55:39 +0000 2008-03-31 14:55:39 +0000 test case 2 charset2.css text/css 48 ot QGNoYXJzZXQvKiovICJ1dGYtOCI7CgpwIHsgZm9udC13ZWlnaHQ6IGJvbGRlcjt9 545 2008-03-31 14:56:00 +0000 2008-03-31 14:56:00 +0000 test case 3 charset3.css text/css 48 ot QGNoYXJzZXQgInV0Zi04Ii8qKi87CgpwIHsgZm9udC13ZWlnaHQ6IGJvbGRlcjt9 546 2008-04-03 14:06:45 +0000 2008-04-03 14:06:45 +0000 simple quote for charset -> fails test.css text/css 45 ylafon QGNoYXJzZXQgJ3V0Zi04JzsKCnAgeyBmb250LXdlaWdodDogYm9sZGVyO30K