W3C home > Mailing lists > Public > www-validator-css@w3.org > August 2004

Re: Two CSS Validator Suggestions

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Tue, 17 Aug 2004 22:45:27 +0300 (EEST)
To: Texas Crazy <howdy@texascrazy.com>
Cc: www-validator-css@w3.org
Message-ID: <Pine.GSO.4.58.0408172231300.1011@korppi.cs.tut.fi>

On Tue, 17 Aug 2004, Texas Crazy wrote:

> When I ran the validator a "warning came up with a message that is confusing
> and may be taken in two opposing ways:
> a.. Line : 0 font-family: You are encouraged to offer a generic family as a
> last alternative

The content of that suggestion is so debatable that I would suggest that
this particular test be removed. It confuses people, especially since most
style sheets contain real errors and problems that need attention, and the
"CSS Validator" does a very useful job in reporting them.

The difference between, say,
font-family: Arial
font-family: Arial, sans-serif
is that the former says "please use Arial if you can; otherwise use
whatever font you were about to use" whereas the latter says
"please use Arial if you can; otherwise use the typical sans-serif
font that you have been configured to use". It is questionable whether the
latter is better even in principle, in general. And in practice, it
invokes both some browser bugs in dealing with generic font names and
some odd choices in browsers - the defaults for their meanings are often
relatively unsuitable fonts, and IE, the most common browser, does not
even let the user change these defaults.

There are situations where a generic family name as the last option
is suitable. But it surely violates no specification to omit it.
And if a program called "validator" issues a warning, I think it is fair
to assume that is should identify a potential _error_, or at least a real
_risk_ (like the risk in setting color without setting background).

> Does this message mean "always ADD a font-family to every text related
> option"?

No, and I have never thought it could be interpreted that way. But maybe
I'm used to reading the messages, where the fake line number reference is
followed by the name of the property that the message relates to.

> Secondly and more importantly, since I am now a bit confused by the warning
> readout, I would suggest a related "details" hyperlink be placed inline in
> the warning section by each error message

That would be nice if done properly, but I'm afraid it's unrealistic to
expect such a development. It means a lot of work, especially since the
information should be concise, understandable, and correct - these three
seldom meet. :-(

Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/
Received on Tuesday, 17 August 2004 19:45:32 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:40:40 UTC