Re: Conformance ratings (was: Extra! Microsoft beats Netscape in the race for...

Daniel W. Connolly (connolly@beach.w3.org)
Sun, 18 Feb 1996 23:30:16 -0500


Message-Id: <m0toNEu-0002U5C@beach.w3.org>
To: Abigail <abigail@tungsten.gn.iaf.nl>
Cc: www-html@w3.org
Subject: Re: Conformance ratings (was: Extra! Microsoft beats Netscape in the race for... 
In-Reply-To: Your message of "Tue, 13 Feb 1996 00:15:45 +0100."
             <199602122315.AAA16878@tungsten.gn.iaf.nl> 
Date: Sun, 18 Feb 1996 23:30:16 -0500
From: "Daniel W. Connolly" <connolly@beach.w3.org>

In message <199602122315.AAA16878@tungsten.gn.iaf.nl>, Abigail writes:
>You, Jonsm@aol.com wrote:
>++ 
>++ I would suggest these categories:
>++ 
>++ 1) Errors - pages containing severe errors like overlapping tags,
>++ quote/comment problems. Any page that can't be parsed by a SGML system gets
>++ this rating. This doesn't mean all of the tags/attributes will be understoo
>d,
>++ it just means that the page is not lexically correct.
>
>That's quite hard to do. While
><img src = foo.gif alt = "bla bla>Text <img src = foo.gif"> or
><!-- -- -- Comment comment -- --> Text text <!-- -- Foo -->
>are most likely "errors", they _are_ correct SGML.


But it's easy to issue warnings in these cases. For example, in
the implementaiton accompanying "A Lexical Analyzer for HTML and Basic SGML,"
at: http://www.w3.org/pub/WWW/MarkUp/SGML/#sgml-lex
I implemented a warning for missing quotes:

excerpt from http://www.w3.org/pub/WWW/MarkUp/SGML/sgml-lex/sgml.l
$Id: sgml.l,v 1.9 1996/02/07 15:32:28 connolly Exp $

  /* <a href = ^http://foo/> -- unquoted literal HACK */
<ATTRVAL>[^ "\t\n>]+{ws}        {
                                  ERROR(SGML_ERROR,
                                        "attribute value needs quotes",
                                        yytext, yyleng);
                                  ADD(SGML_LITERAL, yytext, yyleng);
                                  BEGIN(ATTR);
                                }


It would be simple enough to add warnings for the case of > within
attribute value literals and comment declarations with multiple
comments (e.g. <!-- xxx -- -- yyy -->)


Dan