Next message: Hugo Haas: "Re: checklink.pl fails: missing sys/syscall.ph?"
Date: Wed, 27 Jun 2001 17:34:56 +0200
From: Terje Bless <link@tss.no>
To: Karl Dubost <karl@w3.org>
cc: www-validator@w3.org
Message-ID: <20010627175626-b01010705-dc98314b-0910-010c@192.168.1.6>
Subject: Re: validator feedback
On 26.06.01 at 15:05, Karl Dubost <karl@w3.org> wrote:
>At 19:06 +0200 2001-06-25, Terje Bless wrote:
>>>It will be possible to do it for HTML validator if we can control the
>>>output.
>>
>>For every version and variant from HTML 2.0, through XHTML, to MathML?
>
>Yes it's just a question of time and modularity... and people
>participating in the documentation.
Definitely! If it weren't for Scott Bigham's contributions we wouldn't have
any docs at all.
>If you're looking at the CSS validator, there are french, german, chinese,
>japanese version etc.
Localized versions are not the main issue.
>The problem is not a problem of volume but more a problem of architecture.
Yes, exactly. I don't think we can get enough information out of our SGML
Parser to facilitate this. The best we can do is to have links from the
generic error explanations to relevant parts of as many specs as possible.
The error explanations may of course be inline so this is presented along
with the error message, but I don't think we can get to the point where we
say "This violates section m.n of the FooML 1.0 spec.". We could have
gotten fairly close for just a single spec -- HTML 4.01, say -- but I don't
think it will be feasible to do as a generic mechanism.
OTOH, I've been wrong more then my fair share lately so it's entirely
possible that I'm just missing something glaringly obvious. :-)