W3C home > Mailing lists > Public > www-svg@w3.org > January 2006

RE: Proposal for unitless lengths (Was: Is the px unit equivalent to user units?)

From: Doug Schepers <doug.schepers@vectoreal.com>
Date: Thu, 5 Jan 2006 03:09:19 -0500
To: "'Anne van Kesteren'" <fora@annevankesteren.nl>, "'Jon Ferraiolo'" <jonf@adobe.com>
Cc: "'Jonathan Watt'" <jonathan.watt@strath.ac.uk>, <www-svg@w3.org>, "'Dean Jackson'" <dino@w3.org>, "'Chris Lilley'" <chris@w3.org>
Message-Id: <20060105080923.CAAD5195F6F@plunder.dreamhost.com>

Hi, Anne-

Anne van Kesteren wrote:
| Quoting Jon Ferraiolo <jonf@adobe.com>:
| > (1) One key thing to keep in mind is that there is high 
| > desirability to have a single stylesheet engine which
| > knows how to deal with 'font', 'font-size', 'letter-spacing' 
| > and 'word-spacing' on a universal basis.
| How is this possible given that CSS does not have scientific 
| number notation etc. as noted elsewhere in this thread?

Excellent point, Anne. I think it is important to address any possible
issues with CSS so that implementors (particularly those with existing CSS
engines) have as clear a path as possible to implement SVG in a conforming

It's stated in SVG 1.1, section 4.1 (Basic data types) [1], under the
definition for <number>, "CSS2 [CSS2] states that a property  value which is
a <number> is specified in decimal notation. [...] Thus, for conformance
with CSS2, any property in SVG which accepts <number> values is specified in
decimal notation only." While, "For SVG's XML attributes, to provide as much
scalability in numeric values as possible, real number values can be
provided either in decimal notation or in scientific notation". And further,
under the definition for <length>, that "Note that the specification of a
<number> is different for property values than for XML attribute values."

On the surface, this says to me that only "non-style" (or geometric)
attributes can have scientific notation as a value, removing any conflict.
In pragmatic terms, however, implementors might be tempted to loosen that
distinction, creating a conflict with CSS parsers since all XML attributes
with corresponding style properties must apply the value to the property as
the computed value.

I suggest that to resolve this issue, we should apply the same logic that
Dean suggested in resolving the issue with units. The application of the
value for unitlesss <number>s to unitful <length>s would seem to require an
explicit "casting" of the attribute value, so similarly, we should require
the casting of a scientific notation in any offending style attribute to the
equivalent decimal representation in the computed property. However, this
should not affect the attribute value literal itself. Obviously, how the
rendering engine itself uses the presentation value is
implementation-specific (it may use octal for all it matters [2]). Of
course, in CSS style properties (inline or in a sheet), the decimal notation
should be required.

Does this seem acceptable?

[1] http://www.w3.org/TR/SVG/types.html
[2] Why do programmers confuse Halloween with Christmas? Because oct-31 is


www.vectoreal.com ...for scalable solutions.
Received on Thursday, 5 January 2006 08:09:36 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 March 2017 09:47:06 UTC