Re: [1.2T-LC] attributeType auto (ISSUE-2082)

Anthony Grasso:
> Hi Dr. Hoffmann,
>
> The SVG Working Group discussed ISSUE-2082 and we agree that it seems to be
> underspecified and needs to be addressed.
>
> As you know, however, CSS styling is not required for SVG Tiny 1.2 and the
> behaviour of implementations regarding this attribute is currently unknown.
>
> Would you be comfortable with this issue being postponed and addressed in
> the SVG Core time frame where it will be more relevant?
>

Hello,

there is no need to solve the problem completely already in SVG1.2,
but then there are still some minor adjustments to do to avoid that
something nasty happens, if an implementor really starts to implement,
what is currently defined for "auto" ;o)

Up to now I do not really understand, what might be intended by the
change in SVGT1.2 compared to SVG1.1 or SMIL, therefore I cannot
even suggest something, how to avoid problems.

In the styling chapter it is noted:
"Authors must not rely on external, author stylesheets to style documents that 
are intended to be used with SVG Tiny 1.2 user agents."

For me this does not indicate, that CSS styling itself is not required 
for user-agents or to interprete attributeType="CSS" or 
attributeType="auto" correctly, because this is no external styling. 
Because there is no style element or attribute and authors must not
rely on external stylesheets, there is just no other way for authors
as to use the presentation attributes - or to use attributeTye="CSS"
to apply properties as properties. As already discussed several times,
these priority cascade in SMIL might be unneccessarily complicate
and may be historically not intended, however, because there is
no rule saying something different, this animation type is applicable, 
not only optional. 

One way to avoid confusion could be to leave it as it is defined in
SVG1.1.
Another way could be to say explictly, that the support of attributeType="CSS"
and attributeType="auto" is optional too (this causes a problem, because
"auto" is the default, therefore authors have to be recommended to set
always attributeType="XML" explicitly, if they need to have predictable
behaviour).
A third way could be to note clearly, what is really intended and what 
authors can expect, if they use the default or set 
explictely attributeType="auto" or attributeType="CSS".
Another way could be to remove the values "CSS" and "auto" completely
and to say, that animation in SVG1.2 applies always to the (presentation) 
attributes and to nothing else.

All these variants are not really a problem, if it results in a somehow
predictable behaviour at least for the default value or authors are
forced to set the value explictly to "XML".

Because for SVG1.1 it was typically not very important to care about
attributeType and in SVGT1.1 there was no need to care about it
at all, this new definition of "auto" implicates the opposite case, that
authors of SVGT1.2 have to pay much attention about this to get
a predictable effect. If this is really intended, I think, a big fat warning
note is required to change the typical behaviour of authors completely
for SVGT1.2 and to ensure, that user-agents like Opera really change
their behaviour for SVGT1.2 to this less convenient new rules.


Olaf

Received on Monday, 13 October 2008 11:43:31 UTC