Re: no-namespace href in SVG

Hi, Maciej-

Maciej Stachowiak wrote (on 10/11/2007 7:02 PM):
> 
> On Oct 11, 2007, at 2:48 PM, Doug Schepers wrote:
> 
> I would guess Firefox 2 alone has orders of magnitude more market share 
> than ASV. This survey shows it at ~13% use share: 
> <http://marketshare.hitslink.com/report.aspx?qprid=6>. I would be 
> surprised if ASV has even a tenth this much penetration of browser 
> users, but it's hard to tell since plugins are not reflected in the UA 
> string. 

You may be right.  I have no reason to doubt your source, since I'm sure 
you, working for a major browser vendor, keep pretty close tabs on such 
things.  But when I look at that pie chart, I see the 2 huge slices 
representing IE6 & 7 (triple and double FF's share, respectively, and 
combine to more than 5 times FF).  Assuming any increase in SVG's 
popularity, some portion of that 77% will still rely on ASV (for the 
present).  I don't want to force them to change browsers just to view 
SVG (since SVG would probably lose the battle of momentum).

But you're ignoring the fact that FF is more strict about namespaces 
than ASV!  It also requires an XLink namespace declaration, in addition 
to the attribute prefix.  So, really, over 90% of the browser market 
share is composed of UAs that require the 'xlink:' prefix (directly or 
indirectly).  Maybe FF could change, but we're not hearing from any 
decision-makers there about this issue, yet; they are pretty strict with 
XML, so I'd be interested to hear their take.


> Data from SVG-oriented sites would probably  be
> disproportionately biased to users who are aware of and care about SVG, 
> so I'm not sure it is a useful metric.

It's useful to me, since I care about the experience of people who care 
about SVG.  I don't welcome the undue risk of breaking new content for 
those who are invested in SVG.  SVG is deployed on many intranets, and 
those users typically standardize on IE and ASV... I don't want to give 
them the impression that the "new version" of SVG is incompatible with 
their existing infrastructure, for risk of losing that vertical to some 
proprietary replacement.


>> Opera is very rapidly catching up in performance (if it hasn't already 
>> caught up), and has passed it in lots of very cool features 
>> (especially in Opera 9.5)... they are kicking serious ass.  Both 
>> Firefox and Safari are showing rapid improvement in both categories as 
>> well, though both still lack declarative animation and SVG fonts (IIRC).
> 
> Results that I have seen show Safari 3 to be the fastest implementation 
> for many cases, particularly dynamic script-driven scenarios. For 
> instance <http://data.xeoh.net/svg.benchmark/> (note that higher numbers 
> are better on this benchmark), <http://www.ajaxperformance.com/?p=58>, 
> <http://lists.macosforge.org/pipermail/webkit-dev/2007-April/001772.html>.

I stand corrected.  Kick ass!  I don't use Safari much (might change 
later this month when I get a MacBook), so I was going on my last 
impressions.  Note that in that email, though Andreas reports FF to be 
slow on the Mac, he later tested it a newer build and found the 
performance greatly increased.  So it seems like all 3 browsers are 
doing very well. :)


> I'd agree that Opera's implementation is more complete in most areas.

Some of the experimental stuff is particularly exciting.


>> I don't know if native SVG support in the other browsers will put 
>> significant market pressure on IE to implement SVG, but it may.  At 
>> that point, with active development on all major browsers, there will 
>> be breathing room to consider enhancements like this.
> 
> I'm not sure why that would have to be a blocker, we should not hold 
> back solely because of a low and shriking market share abandoned 
> implementation.

Were that the only argument against allowing null-namespace 'href', and 
were the evidence stronger that it would help authors significantly 
without causing commensurate confusion, I would be more swayed.  This 
thread has been pared down to a discussion of ASV's dwindling market 
share, but that's not the big picture.

I reckon that this is a battle in a larger war against namespaces.  I 
understand that you don't like namespaces, but in my experience, it 
takes authors only a few minutes to learn about them, and then they have 
another tool in their belt.  I am not interested in engaging in an 
ideological fight about it (right now :) )... my main interest is in 
what is best for authors and for SVG.  I seen benefits and challenges on 
both sides of the argument, but no argument has yet convinced me of the 
necessity of changing something that already works, even if it's not 
perfect.  But I'm only one person in the SVG WG, so ultimately, it's not 
up to me; it's got to be a group decision.


> Currently it seems like libraries for structured vector graphics on the 
> web tend to use SVG in browsers with native support and VML in IE, 
> regardless of whether ASV is available. Hard to tell what the long-term 
> story for IE is.

Agreed.  Pity we need those libs, which degrade to the lowest common 
denominator in terms of graphics formats (though they also offer nice 
helper functions that SVG and JS should learn from).  Hey, IE Team... 
people want SVG!


Regards-
-Doug Schepers
W3C Staff Contact, SVG, CDF, and WebAPI

Received on Friday, 12 October 2007 05:27:04 UTC