SVG complexity, plus predictions

Ok, I'm going to put in my two cents.

The single biggest problem with SVG is its overweening complexity. The
spec itself is complex enough, but the sheer volume of stuff it includes
by reference is starting to be demoralizing for me. Things seem to be
getting worse, not better. For example, why are there two styling
languages? Styling itself seems like a feature of limited use within
SVG, as SVG is primarily a presentational form without real text
editability.

Here are my predictions:

1. People will join in the handwringing inspired by this post, but at
the end of the day, no significant fat will be cut from the spec.

2. When the SVG spec is published as a recommendation (I hesitate to say
"finalized" because I can hear people already licking their chops in
preparation for a 2.0), only Adobe will have a near-complete
implementation. It won't be 100%, though, as that's just too much to
ask.

3. Further, the spec will not nail down all interoperability problems.
In particular, it will wink and nod on SVG generators that do not
include the Bezier outlines of all fonts used.

4. The various SVG implementations will have severe interoperability
problems, due primarily to nobody implementing all the features, and
also due to the font problem. Adobe's SVG will become the de facto
standard. OpenType will become the de facto font standard, as Adobe will
at least include the option of linking an OpenType font rather than
including all the shapes in the SVG file.

5. Adobe Acrobat 5, implementing PDF 1.4, will actually ship in
mid-summer, probably well in advance of the 1.0 release of the SVG spec.

6. PDF 1.4 will steamroller SVG. It's a much simpler format, is familiar
and has the perception of a lot of momentum in industry, and from what I
can tell will have an imaging model that surpasses SVG's. PDF 1.3 is
already more advanced in some ways, especially its gradient mesh
capabilities. The main thing lacking is transparency, and it seems
likely that's going to happen. PDF 1.4 will also work well for print,
basically precluding the possibility of SVG ever gaining a foothold in
that area.

7. Mozilla is going to take a while before it gets a usable SVG
implementation. I have no idea about Microsoft. The upshot is that
without an installed base of clients, sites won't be putting a lot of
SVG up any time soon.

8. Yet, SVG will gain a definite niche. Its main advantage is that it's
quite easy to generate, a feature <i>not</i> shared by PDF. It will
become a favorite for websites that generate graphics dynamically,
especially from XML sources. These graphics will generally be delivered
to clients as GIF and JPEG due to the unreliability of content
negotiation.

I would love to be proved wrong on any of these predictions (well,
except for number 8), but let's wait a few months and see.

Raph

Received on Wednesday, 8 March 2000 01:59:04 UTC