Re: [SVGMobile12] more on data types

On 1/11/06, Jim Ley <jim@jibbering.com> wrote:
>
>
> <mental@rydia.net> wrote in message
> news:1137005288.43c552e829753@www.rydia.net...
>
> >Quoting Christophe Jolif <cjolif@ilog.fr>:
> >
> >> If not, I would be afraid that same SVG content would lead to
> >> different rendering on different UA.
> >>
> >> Imagine: fill="#fff 111"
> >>
> >> on can recover it as #fff
> >>
> >> and the other one as #fff111
> >>
> >> which is not exactly the same...
> >
> >If it's not conformant, should we care?
>
> Indeed, that is my point, Working Groups should care about conformant
> content, and how to process it, if they want the above to be consistently
> implemented by user agents, then all they have to do is make it conformant
> and specify the algorithm for determining the colour.
>
> The options for the working group are Must ignore, which means specify
> everything they want to be implemented identically as valid, and everything
> else be ignored, or undefined it's up to the implementation what they do -
> that is far my preferred option, must ignore is not useful to users.
>
> Cheers,
>
> Jim.
>

I firmly believe that SVG viewers should be required to stop and
display an error message whenever they come across and error, and that
they should check to see that the content is valid.   However, it
isn't the specs job to be dictating algorithms.  The meaning and form
of the data is what should be defined in the spec, implementors should
be allowed to create their applications as they see fit, as long as
they produce UA's that behave in accordance with the spec.

When I author content, I test it continually in every viewer and OS
that I can.  I have two viewers on Linux and four on Windows.  I want
my stuff to be correct and portable.  And I can't type.

FWIW, Batik (Squiggle) is the best aid here as far as error
information is concerned, but getting all the error messages helps,
and for the UA's that don't give error messages, at least I can look
at the image and see that it is what I am expecting.

Most content authors will probably not do this.  They will be
authoring to a specific viewer and not be concerned, or possibly even
aware of the UA's which may be viewing their content when it is
published.  Nor should they have to be.

I've implemented SVG viewers, it isn't difficult to check to see that
attributes that aren't namespaced are valid, and it isn't difficult to
check to see that attribute values are correct.  Especially on desktop
applications.

It's easy to say that it isn't the viewers responsiblilty to check
content, and there are good arguments for that.

What is difficult is dealing with erroneous content, not just as an
author, or as an individual, but as a community.  We should do
everything we can, as implementors, WG members, and SVG enthusiasts,
to assist authors in developing proper content, and if they are
allowed to produce erroneous content because the tools that they have
available allow it, we cannot fault them and we will deserve the
legacy that they will leave behind.

--
Cheers!
Rick

Received on Wednesday, 11 January 2006 19:53:41 UTC