W3C home > Mailing lists > Public > www-svg@w3.org > April 2003

RE: [SVG] Experiment and proposal

From: Fred P. <fprog26@hotmail.com>
Date: Sat, 26 Apr 2003 20:24:54 -0400
To: doug@schepers.cc, www-svg@w3.org, svg-developers@yahoogroups.com
Message-ID: <BAY2-F24amg6dfEm58z00001ff0@hotmail.com>

Thanks Doug for your diplomatic approach!

>Hi, Fred-
>First, may I make a suggestion? You seem to be talking about several
>different issues within the body of one very long email, in a somewhat
>scrambled order. This makes your arguments a bit hard to follow at times.
>For the sake of clarity, you might want to split your posts into several 
>different threads:
>1) the possiblility and specifics of implementing layout grids and/or
>native/XHTML-namespace tables into SVG (in fact, these might well be 2
>different topics);
>2) the idea of allowing the insertion of raster images via markup pixels, 
>rather than as binary data;

I gave up on that one (2), but I started a new thread for (3),
for proper rendering of polygon in aliased mode.

>3) the problem of 1-pixel variation of very small images using aliasing;
>4) text issues;
>5) some other things?
>As I said, this is only a suggestion, but each issue would stand a better
>chance of not being lost in the shuffle.

I agree. =)

>That being said, let me address points 2 and 4.
>2) I don't quite understand why you would want to do this. I would be very 
>surprised if a series of dots described in standard SVG syntax, such as:
>    <circle cx='25' cy='25' r='1' style='fill:orange; '/>
>or even:
>    <dot cx='25' cy='25' style='fill:orange; '/>

I did some test and found out it looks bad, so I gave up on that one.

>would come at all close, even if gzipped, to that of the advanced raster
>compression schemes. And even were that to be the case, it seems an odd use
>of vector graphics.

Well, it is the case, believe it or not. =P

271KB of PNG/GIF = 1.5MB of SVG = 71KB of SVGZ

As far as I'm concern, I don't use SVG for vector graphics,
I used SVG because it's W3C, it has plug-ins and it is better
supported than GXL or similar.

Secondly, creating graphics from a database is more feasible
and easier if the end result is in XML format compared to GIF,PNG,BMP,JPEG 
or similar. There is some text format
like XBM, XPM and similar but there is
no plug-in for it on Windows, for instance.

Another reason for using SVG is because it support dynamic content
via ECMAscript or JavaScript, which is a well-known language.
and it's quite easy to do and surprisingly Adobe support
quite a BLAZING FAST renderer.

To tell you a little story, I was searching for a way to display
UML class diagram on the web without having a SLOW java applet
or DHTML problems with Netscape. By mistake, I found a book on
SVG and found out it was a W3C standard that was far more
supported and had more feature than GXL which was popular few years ago,
some IEEE paper, like the one on Rigi/SVG starts to spread the word
that SVG is really cool and should be used to render UML diagrams
and similar.

My purpose is that UML diagrams have to be PRECISE at the pixel level,
the text must be crystal clear, it must be easy to display text,
to align text, to create 'table' of text with little images at
the left of it and to be able to manipulate this entire thing
with ease with JavaScript support with the maximum amount of
accessibility and usability, like a PROFESSIONAL CAD tools,
like Rational Rose or Visio or similar, all on the web client side,
with blazing performance.

I was able to show you folks some prototype that indeed it is feasible
but there was concerns about getting those nice UML diagram into
vectorialized form instead of pixelized form as demonstrated
in the other thread.

It's not because I use SVG for something that was not planned for
in the first place, that it is therefore not an issue,
extending a cool format like SVG to cover a broader
range of application is something that should be considered.

For instance, having blazing fast Windows-like application
directly on the Web rendered with SVG on a web-browser
would be something quite cool isn't it?

It's already possible, it's already very efficient.
Now we just have to fix the glitches to make it
truely professional and widespread.

>I looked at your examples, particularly the ones with
>icons for folders, locks, etc., and when zoomed in, they look just as I
>would expect a raster to look when enlarged --that is, pixelated and
>blocky-- which is not at all what I expect from vector graphics --which
>should be smooth and look like smaller or larger version of the same icon.

Exactly, but if you see the hand example, the path version
scales well, but look terribly bad at low resolution,
more specifically at pixel level.

>No offense, but I don't see what the use for this is. I would think one 
>should draw new icons that leverage the abilities of SVG, rather than try 
>reuse old bitmap icons.

Because those icons looks great. Why change them?
My only concerns would be to make them vectorialized,
but it's not possible, since we lose quality.

>Maybe if you explained more thoroughly what exactly you are trying to
>achieve with this, it would seem more applicable to SVG.

A true web application that is 100% identical at pixel-level
to a Windows-like application, but that would scale
since it would have vectorialized graphics, currently
this is a plus-value, not a requirement, but it would be nice to have.

>4) On the subject of text selction, I think I can be of more help. If I
>understand you correctly, you want to disable mouse-based text >selection.
>You can do this easily:
>    <text x='20' y='13' style='pointer-events:none;'>Unselectable 

Wow, thanks!!!
That works great.
But I might get a hard time calculating the size of the transparent
rectangle. However, since the text is within an object, I might be
able to go around this. I'm looking into it right now.

Thanks doug! =]

>There are a few problems with this method --the text can no longer be
>selected as a target for any mouse events, such as click or mouseover. 
> >There
>are actually a number of different methods involving JavaScript to overcome
>this limitation (for example, if you want to grab and drag the text, you
>might have an "invisible" rectangle around the text that does respond to
>mouse events), but I think there could be a better solution. I mentioned it
>on this list before (as has Jim, I believe):
>By the way, if you have questions about how to achieve different
>implementation tasks using SVG (like, how to turn off text selection), you
>would be better off posting to svg-developers@yahoogroups.com. The W3C SVG
>list is more for proposing changes or additions to the SVG Spec (such as
>calling for the introduction of grids or tables into SVG).
> > Still, having a CSS stylesheet with > style="text-selection: disabled;" 
>would be a good thing,
> > for buttons, menus and SVG widgets.

>Hope that helps-

Sure it does!

Sincerely yours,


Add photos to your e-mail with MSN 8. Get 2 months FREE*.  
Received on Saturday, 26 April 2003 20:25:01 GMT

This archive was generated by hypermail 2.3.1 : Friday, 8 March 2013 15:54:24 GMT