W3C home > Mailing lists > Public > html-tidy@w3.org > October to December 2001

Re: CheckColor default

From: Klaus Johannes Rusch <KlausRusch@atmedia.net>
Date: Sun, 4 Nov 2001 16:56:45 CET
Message-Id: <200111041625.LAA07572@tux.w3.org>
To: html-tidy@w3.org
In <migautgds932fv2iqseb9fge4kflds8lu3@4ax.com>, Bjoern Hoehrmann <derhoermi@gmx.net> writes:
> * Klaus Johannes Rusch wrote:
> >In <jgaautg3267nbnq1e5qsth3k4rnfqvobrq@4ax.com>, Bjoern Hoehrmann <derhoermi@gmx.net> writes:
> >> No, the behaivour is intended, Tidy should IMO always transform the case
> >> of hex digits and replace those 16 colors with their corresponding color
> >> names.
> >
> >There are at least two good reasons:
> >
> >* Color names are supported by MOST browsers but there are some, admittedly
> >  historic, browsers which only understand color values.
> Please name them. Color names were included in all HTML versions, ...

Color names introduced in HTML 3.2 but implemented by browser vendors before
the spec was finalized.

Netscape 1.1(!) added support for hex values but not for names

Netscape 1.22 interprets all color names as blue

IBM WebExplorer 1.2 has color support for hex values but not for names.

> ... user
> agents that do not support them are simply non-conforming. There may be
> many that do not support e.g. color=orange, since orange is no valid
> color keyword.

No doubt user agents should have supported color names. They don't.

Following the principle that "an implementation must be conservative in
its sending behaviour, and liberal in its receiving behaviour" (RFC 791
Internet Protocol) choosing something all browsers understand, whether
or not they comply with standards, would be reasonable.

> >* tidy should change what needs to be changed, but not mess with code that is
> >  already valid. The author of a document may have had a good reason, or just
> >  a personal preference, for choosing a color value instead of a color name.
> Tidy generates a canonical version of the document. What this includes
> is not specified.

IMHO it should be specified what the canonical version of a document is, 
otherwise running tidy multiple times against the same input document will 
result in a different output document every time (which makes it absolutely 
useless in an environment with version control because you end up checking in 
modifications which are just tidy's mood of the day).

> >> >The option probably should allow conversions in both directions, i.e.
> >> >
> >> >        ColorAttributes: symbolic|hexadecimal|asis
> >>
> >> I don't see any good reason why one wants Tidy to turn readable color
> >> names to unreadable color codes.
> >
> >I do not see a benefit in making tidy unusable for folks who need to or
> >want to use color values.
> I don't argue about an option to prevent the change the color values, I
> argue about conversions in both directions.


Klaus Johannes Rusch
Received on Sunday, 4 November 2001 11:25:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:38:51 UTC