Re: transform as a presentation attribute

Strictly following the recommendations for SVG 1.0, SVG 1.1, SVG tiny 1.2,
properties not explicitly mentioned in the recommendations are not applicable
for those documents anyway.
For example there are currently no properties width, height, transform etc
applicable for SVG elements. If viewers ignore the recommendations, they
clearly have a bug ;o)
Therefore whatever is changed for SVG 2 concering attributes and properties
or specifities/priorities, this should have no effect on current documents 
indicating to be written in one of the versions 1.x.
Of course we know, that viewers tend to have lots of bugs and gaps and
do not necessarily care about recommendations, documents have been
written with, therefore we can be almost sure, that they will fail in the 
future as well, if there are backwards incompatible changes.
Therefore SVG 2 has the requirement to avoid such backwards incompatible
changes.
This should typically mean, the working group should chose new names for new
properties, which are not in conflict with old attribute names.
But of course, since HTML5 drafts appeared, it is modern now for other
working groups to create tag soup as well - we will see, how much this
will apply for SVG 2.
We know as well, that some implementors intentionally implement bugs
or keep gaps open to corrupt formats - but there is not much we can
do as to prefer other programs or to continue to send bug reports and
explaining the purpose of recommendations.

I think as well, that there is not much need to convert attributes like 
transform and many others to properties, because typically they are only
related to content and not to styling.
On the other hand it can be a simplification for authors, to be able to use
CSS, if decorative alternatives really require transforms. Many are not
very familiar with XSL(T) to apply this instead of CSS, if they really have
to modify elements or attributes for styling reasons.
And my assumption is, if an author really applies a stylesheet to an SVG
document or a mixtures of SVG and XHTML, the risk is pretty low, that
accidently new CSS features are applied to SVG documents of version 1.x
Why should authors do this? Typically authors know as well, that most
viewers have no version specific implementations and fail therefore for
backwards incompatible changes in recommendations, therefore they will
simply not add new features to old documents or  to old stylesheets without 
checking carefully for such viewer bugs. 

And for SVG 2 documents they should know, that attributes converted to
properties are changeable due to stylesheets, therefore no problem to be
selective enough to avoid unintended conflicts.

I think, often it is not very useful to change attributes to properties, but 
sometimes it has usecases and the risk of nonesense
presentation for old files is low. But there is a big risk, that due to the
abuse of stylesheets, there will be no meaningful primary presentation
(without CSS interpretation) anymore, if programs like inkscape insist
to use the style attribute for everything, that can be expressed as a 
property. Such abuse of the style attribute for nonesense is already today
a much bigger risk for the audience to get inaccessible, meaningless 
documents. 
Therefore clearly one should not promote to use the style attribute
at all or something like this with similar high specifity/priority.
Instead it is important to explain authors, that they always have to
clean up files produced by programs like inkscape to get rid of
style attributes in published documents.

Changing the specifity/priorities of presentation attributes within the
CSS cascade together with the known bug of most viewers not to care
about format versions indicated in documents, means always, that 
most future viewers will fail for old documents, which use external CSS files
to provide alternative views with different styling.
Therefore in practice your suggestion causes trouble for old documents
if implemented in those viewers with versionless interpretations
- the opposite of your intentions. In current documents there is no need
for the !important rule, fortunately! And hopefully in the future there will
be still no need to use this - this is mainly a feature, if already something
went wrong, this should not be the normal way to provide styling for SVG 
documents in the future.


Olaf

Received on Tuesday, 24 June 2014 16:03:44 UTC