In the wake of XTech 2000, XML-DEV members have been
taking an analytical look at SVG, concerned in particular by
the use of attributes to contain extended path information.
Don Park (who else?) kicked
off the questioning,
observing that the long list of path instructions in
attributes prevented structural manipulation via common XML
processing tools such as the DOM or XSLT.
Peter Murray-Rust sympathized
with some of the design decisions the SVG had to take,
describing his own experience with CML. However, he stated
his desire to be able to perform XSLT-based parsing of SVG.
(It was later pointed out that the SAXON[?] XSLT processor's
tokenize() extension function may be of use here).
Jon Ferraiolo of Adobe, editor of the SVG spec, posted a
response to the design queries, citing four points
in favor of the attribute notation for path data:
- The desire to keep SVG file size down
- The largest contributor to file size is path data
- Compression on verbose path syntaxes didn't yield much
additional benefit than compression on compact syntaxes
- The DOM could become huge if each path data command was
its own element
Ferraiolo then went on to expand on the design
considerations, and commented:
We really appreciate everyone's interest and we like to hear
feedback of
any type. Even if we don't respond to feedback, we are
monitoring
www-svg@w3.org and discuss the topics that come up there in
working group
discussions.