misc.txt

download original
15:10 < multi_io> and as I said, I don't think it solves my problem. My problem is this: I get
          a mouse click/move event on an SVG element. The coordinates in the event
          (clientX/clientY) are relative to the browser window (or, in the case of
          layerX/layerY, relative to the rendered HTML page). I want them to be relative to
          the SVG element the event occured on. So I need the upper-left corner of that
          element in the page. How do I obtain that?


15:44 < raxor> multi_io: still around
15:45 < raxor> ?
15:45 < multi_io> raxor: yeah
15:45 < raxor> getBBox is not in Firefox, you are right
15:45 < raxor> There are two ways of doing it
15:45 < raxor> one is to think about the getTransformToElement...
15:46 < raxor> the other is to think about the getScreenCTM
15:46 < tor> getBBox is definitely in firefox
15:46 < raxor> gah
15:46 < raxor> sorry
15:46 < raxor> I mean... it is not in firefox onload!
15:46 < tor> you can't currently use it at onload time, but it should work in other
             circumstances
15:47 < raxor> tor is always right on these things multi_io :)
15:47 < raxor> tor is getScreenCTM in FF?
15:47 < raxor> or getCTM?
15:48 < raxor> multi_io: my guess is that you are trying to do something like drag a box
               around a box or figure out the coords relative to a container for mapping...
               right?
15:48 < tor> yes to both (on trunk at least, don't have a branch tree around to check)
15:48 < tor> ok, I've just made ff trunk pickier about the transform attribute
15:48 < multi_io> raxor: a freehand drawing application, sort of
15:49 < raxor> okay... are you expecting to have nested transforms?
15:49 < raxor> e.g., <svg> within <svg>, or <g transform="....">
15:49 < multi_io> yes, probably.
15:49 < raxor> then it gets really fun :)
15:50 < multi_io> I'm not scared of matrices or anything
15:50 < raxor> Good good... it is more the variance in implementations you should be worried
               about :)
15:50 < multi_io> hm yeah, I think I'm getting a glimpse of that already :)
15:51 < raxor> It is better now...
15:51 < raxor> but when you have ASV3 in the mix it is tough
15:51 < multi_io> incidentally, all the freehand apps I could find on the web get it wrong too
                  -- see http://lamp2.fh-stpoelten.ac.at/~lbz/beispiele/ws2006/draw/index.xml
15:52 < raxor> I ended up writing a whole js layer for coord transforms so I could eliminate
               those idiosyncracies
15:52 < raxor> there is still some debate over the "correct" implementation... so I chose my
               theory
15:52 < raxor> http://research.pdmagic.com/svg/
15:52 < raxor> http://research.pdmagic.com/svg/coords.js
15:54 < multi_io> thanks raxor
15:54 < raxor> Basically you need to override var svgRoot = null; var svgDoc = null; var
               svgWindow = null; then you can call getScreenCTM
15:55 < raxor> In any event... looking at the code (for getScreenCTM) will give you an idea of
               how to translate the coords...
15:55 < raxor> if it is FF only, I think you can just use their built in function
15:56 < raxor> let me know what you go with and I can try to help some more...
15:57 < multi_io> ok
back to drawing