Loading ...
Sorry, an error occurred while loading the content.

Re: [svg-developers] Re: Swapping attribute values on large numbers of elements

Expand Messages
  • Andre M. Winter - Carto.net
    hi, in a situation like this: i expected to get a mouseup-event only when finishing dragging. but
    Message 1 of 23 , Feb 1, 2007
    View Source
    • 0 Attachment
      hi,

      in a situation like this:

      <svg ... onmouseup="myDrag(evt)" onclick="myClick(evt)".../>

      i expected to get a mouseup-event only when finishing dragging. but i
      get a click-event too, no matter how long i dragged my stuff around.
      ASV, FF and Opera do so, hence i expect this to be the clean behavior.
      but how differentiating those two events? think about an overview map
      where dragging the extent is possible and also clicking in order to set
      a new one...

      thx
      andré

      --
      ___________________________________________________________________
      andre m. winter,
      cartography for internet and multimedia applications
      schiessstand 4/1, a6091 goetzens, tyrol, austria
      tel.: ++43.5234.32732
      email: ml.winter@...

      new svg book with actual scripting samples out now!
      check http://svg.carto.net/
    • "André M. Winter - Carto.net"
      hi, i was just testing a map interface with FF2 and was disappointed while unloading a map and reloading another form the local filesystem. it took around
      Message 2 of 23 , Feb 13, 2007
      View Source
      • 0 Attachment
        hi,

        i was just testing a map interface with FF2 and was disappointed while
        unloading a map and reloading another form the local filesystem. it took
        around 15sec. now FF3 (feb. 8th) does it in around 1sec and this over
        the web. this is faster than IE6/ASV3. mouse interactions work smooth,
        pages (also html) load at an amazing speed :-)

        give it a try with your own files, here you go:
        http://www.mozilla.org/projects/firefox/3.0a2/releasenotes/#download

        andré

        --
        ___________________________________________________________________
        andre m. winter,
        cartography for internet and multimedia applications
        schiessstand 4/1, a6091 goetzens, tyrol, austria
        tel.: ++43.5234.32732
        email: <winter@...>

        <http://www.vectoreal.com/> SVG consulting and development
        <http://www.geotrace.net/> geo-localized high quality photographs
        <http://www.carto.at/> print and online touristic map solutions
      • T Rowley
        ... Some information about the SVG capabilities in 3.0a2 are listed in the following URL. Note that a number of SVG applications will have problems due to
        Message 3 of 23 , Feb 13, 2007
        View Source
        • 0 Attachment
          On 2/13/07 8:32 AM, André M. Winter - Carto.net wrote:
          > i was just testing a map interface with FF2 and was disappointed while
          > unloading a map and reloading another form the local filesystem. it took
          > around 15sec. now FF3 (feb. 8th) does it in around 1sec and this over
          > the web. this is faster than IE6/ASV3. mouse interactions work smooth,
          > pages (also html) load at an amazing speed :-)
          >
          > give it a try with your own files, here you go:
          > http://www.mozilla.org/projects/firefox/3.0a2/releasenotes/#download

          Some information about the SVG capabilities in 3.0a2 are listed in the
          following URL. Note that a number of SVG applications will have
          problems due to regressions caused by some tree-wide gecko changes. The
          good news is that we now have a handle on these issues and have patches
          in hand to fix them.

          One thing I forgot to mention in the following is that we now have
          incremental loading of XML, so you'll see progressive rendering of SVG
          files as they load.

          http://weblogs.mozillazine.org/tor/archives/2007/02/gran_paradiso_alpha_2_and_svg.html

          -tor
        • Antoine Quint
          Hi there, As an additional note, I d like to say that I get to work every day with the latest (give or take a couple of weeks) trunk of Gecko, which is also
          Message 4 of 23 , Feb 13, 2007
          View Source
          • 0 Attachment
            Hi there,

            As an additional note, I'd like to say that I get to work every day
            with the latest (give or take a couple of weeks) trunk of Gecko,
            which is also used in Mozilla, for Joost and that the progress in SVG
            support and mixing with other XML grammars supported natively in
            Gecko has been truly amazing. Not surprised that André is already
            benefiting from it. Firefox 3.0 should be a significant milestone for
            SVG development, just like FF 1.5 and Opera 9 have been before it.

            Antoine
            --
            Blog — http://the.fuchsia-design.com
          • "André M. Winter - Carto.net"
            hi tor, ... well, i reported my experience from a user perspective. i took a file out of my actual interface development (sorry not public yet) and it worked
            Message 5 of 23 , Feb 13, 2007
            View Source
            • 0 Attachment
              hi tor,

              > Some information about the SVG capabilities in 3.0a2 are listed in the
              > following URL. Note that a number of SVG applications will have
              > problems due to regressions caused by some tree-wide gecko changes.


              well, i reported my experience from a user perspective. i took a file
              out of my actual interface development (sorry not public yet) and it
              worked out of the box. it's not a simple static file (html-frameset with
              overview map, current.scale/translate-based zoom&pan-controls, change of
              the iframe-src of the main map, re-population of the overview (DOM
              manips) and DOM-interaction with the HTML around. and it is ASV3-compliant.

              if such a beast works with no problems in FF3, this is near from water
              proof for me.


              > One thing I forgot to mention in the following is that we now have
              > incremental loading of XML, so you'll see progressive rendering of SVG
              > files as they load.

              didn't notice that yet in practice. but the last years we learned to
              build small small and smaller files, that may be the reason...

              --
              ___________________________________________________________________
              andre m. winter,
              cartography for internet and multimedia applications
              schiessstand 4/1, a6091 goetzens, tyrol, austria
              tel.: ++43.5234.32732
              email: <winter@...>

              <http://www.vectoreal.com/> SVG consulting and development
              <http://www.geotrace.net/> geo-localized high quality photographs
              <http://www.carto.at/> print and online touristic map solutions
            • TheMountainScene
              In a FF test I ve created, a mouseover of Canada 364 Kb, it has improved from 6 seconds to 2 seconds. 2 Seconds is still much too slow compared to the Adobe
              Message 6 of 23 , Feb 13, 2007
              View Source
              • 0 Attachment
                In a FF test I've created, a mouseover of Canada 364 Kb, it has improved
                from 6 seconds to 2 seconds. 2 Seconds is still much too slow compared
                to the Adobe viewer, but it is an improvement. I submitted this as a bug
                some time ago. For my purposes, it still falls short, and given the
                snail's pace anything svg is moving, I see flash in my future. I keep my
                fingers crossed, but my optimism wanes with each passing day.

                Sean

                André M. Winter - Carto.net wrote:
                >
                > hi tor,
                >
                > > Some information about the SVG capabilities in 3.0a2 are listed in the
                > > following URL. Note that a number of SVG applications will have
                > > problems due to regressions caused by some tree-wide gecko changes.
                >
                > well, i reported my experience from a user perspective. i took a file
                > out of my actual interface development (sorry not public yet) and it
                > worked out of the box. it's not a simple static file (html-frameset with
                > overview map, current.scale/translate-based zoom&pan-controls, change of
                > the iframe-src of the main map, re-population of the overview (DOM
                > manips) and DOM-interaction with the HTML around. and it is
                > ASV3-compliant.
                >
                > if such a beast works with no problems in FF3, this is near from water
                > proof for me.
                >
                > > One thing I forgot to mention in the following is that we now have
                > > incremental loading of XML, so you'll see progressive rendering of SVG
                > > files as they load.
                >
                > didn't notice that yet in practice. but the last years we learned to
                > build small small and smaller files, that may be the reason...
                >
                > --
                > __________________________________________________________
                > andre m. winter,
                > cartography for internet and multimedia applications
                > schiessstand 4/1, a6091 goetzens, tyrol, austria
                > tel.: ++43.5234.32732
                > email: <winter@... <mailto:winter%40carto.net>>
                >
                > <http://www.vectoreal.com/ <http://www.vectoreal.com/>> SVG consulting
                > and development
                > <http://www.geotrace.net/ <http://www.geotrace.net/>> geo-localized
                > high quality photographs
                > <http://www.carto.at/ <http://www.carto.at/>> print and online
                > touristic map solutions
                >
                >
              • Guy Morton
                I wish I could report the same happy news. Unfortunately, GranParadiso (on MacOSX at least) fails to render *anything* for me. It even seems buggy in doing
                Message 7 of 23 , Feb 13, 2007
                View Source
                • 0 Attachment
                  I wish I could report the same happy news. Unfortunately,
                  GranParadiso (on MacOSX at least) fails to render *anything* for me.
                  It even seems buggy in doing normalish browser things like resizing
                  framesets (the toolbar has half disappeared under the window title in
                  my testing).

                  Have a look at http://bct2.webtrak-lochard.com/ and see.

                  It's not generating anything in the error logs that gives any
                  indication of what it might be taking exception to either.

                  I hope that whatever is wrong can be ironed out, as having a fast FF
                  implementation would be a real shot in the arm for SVG, whereas a
                  buggy implementation would just about be the kiss of death for SVG as
                  far as this project is concerned.

                  Guy


                  On 14/02/2007, at 1:32 AM, André M. Winter - Carto.net wrote:

                  > hi,
                  >
                  > i was just testing a map interface with FF2 and was disappointed while
                  > unloading a map and reloading another form the local filesystem. it
                  > took
                  > around 15sec. now FF3 (feb. 8th) does it in around 1sec and this over
                  > the web. this is faster than IE6/ASV3. mouse interactions work smooth,
                  > pages (also html) load at an amazing speed :-)
                  >
                  > give it a try with your own files, here you go:
                  > http://www.mozilla.org/projects/firefox/3.0a2/releasenotes/#download
                  >
                  > andré
                  >
                  > --
                  > ___________________________________________________________________
                  > andre m. winter,
                  > cartography for internet and multimedia applications
                  > schiessstand 4/1, a6091 goetzens, tyrol, austria
                  > tel.: ++43.5234.32732
                  > email: <winter@...>
                  >
                  > <http://www.vectoreal.com/> SVG consulting and development
                  > <http://www.geotrace.net/> geo-localized high quality photographs
                  > <http://www.carto.at/> print and online touristic map solutions
                  >
                  >
                  > -----
                  > To unsubscribe send a message to: svg-developers-
                  > unsubscribe@yahoogroups.com
                  > -or-
                  > visit http://groups.yahoo.com/group/svg-developers and click "edit
                  > my membership"
                  > ----
                  > Yahoo! Groups Links
                  >
                  >
                  >
                • T Rowley
                  ... Assuming you re talking about the Canada map in bug 319990, the version you have online right now has a JS error. Correcting that, removing the alert, and
                  Message 8 of 23 , Feb 13, 2007
                  View Source
                  • 0 Attachment
                    On 2/13/07 2:24 PM, TheMountainScene wrote:
                    > In a FF test I've created, a mouseover of Canada 364 Kb, it has improved
                    > from 6 seconds to 2 seconds. 2 Seconds is still much too slow compared
                    > to the Adobe viewer, but it is an improvement. I submitted this as a bug
                    > some time ago. For my purposes, it still falls short, and given the
                    > snail's pace anything svg is moving, I see flash in my future. I keep my
                    > fingers crossed, but my optimism wanes with each passing day.

                    Assuming you're talking about the Canada map in bug 319990, the version
                    you have online right now has a JS error. Correcting that, removing the
                    alert, and moving in/out right near the border (which defeats our
                    early-reject hit detection), the major hotspots in the profile all
                    appear inside the cairo 2D graphics library:

                    57.20% checking if a point is inside the path
                    99.7% of this time is spent tesselating the path to trapezoids

                    14.83% painting paths
                    99% of this time is spent tesselating the path to trapezoids

                    14.51% getting screen path extent
                    99.9% of this time is spend tesselating the path to trapezoids
                    The reason this appears in the profile is because the style change
                    information Gecko gives us is a general "things changed" rather than
                    detail which would let us realize that only the fill changed and the
                    extent didn't need to be recalculated.

                    It seems that if cairo improved their point-in-path functionality using
                    one of the techniques that doesn't need a whole trapezoid list, things
                    would be a fair bit faster.

                    Changing the style system to give detailed information is a larger scope
                    item, though maybe something gecko would want to do as it could help
                    other sorts of content.

                    It does seem as though this map is a bit of a torture case for cairo's
                    tesselator - I've passed on the URL to the cairo developers.

                    -tor
                  • Guy Morton
                    Aha...well that s the problem for me. I resized the window and it appeared. Good to know you know what is causing it. And I ll add my Wow to Andre s...FF
                    Message 9 of 23 , Feb 13, 2007
                    View Source
                    • 0 Attachment
                      Aha...well that's the problem for me. I resized the window and it
                      appeared. Good to know you know what is causing it.

                      And I'll add my "Wow" to Andre's...FF used to use 70% of my CPU
                      playing this file. Now it's using about 7%. There are some screen
                      redrawing issues though still, which you'll see if you look at the
                      link I sent (the planes are leaving translucent colour over the map).

                      Guy




                      On 14/02/2007, at 2:24 AM, T Rowley wrote:

                      > On 2/13/07 8:32 AM, André M. Winter - Carto.net wrote:
                      >> i was just testing a map interface with FF2 and was disappointed
                      >> while
                      >> unloading a map and reloading another form the local filesystem.
                      >> it took
                      >> around 15sec. now FF3 (feb. 8th) does it in around 1sec and this over
                      >> the web. this is faster than IE6/ASV3. mouse interactions work
                      >> smooth,
                      >> pages (also html) load at an amazing speed :-)
                      >>
                      >> give it a try with your own files, here you go:
                      >> http://www.mozilla.org/projects/firefox/3.0a2/releasenotes/#download
                      >
                      > Some information about the SVG capabilities in 3.0a2 are listed in the
                      > following URL. Note that a number of SVG applications will have
                      > problems due to regressions caused by some tree-wide gecko
                      > changes. The
                      > good news is that we now have a handle on these issues and have
                      > patches
                      > in hand to fix them.
                      >
                      > One thing I forgot to mention in the following is that we now have
                      > incremental loading of XML, so you'll see progressive rendering of SVG
                      > files as they load.
                      >
                      > http://weblogs.mozillazine.org/tor/archives/2007/02/
                      > gran_paradiso_alpha_2_and_svg.html
                      >
                      > -tor
                      >
                      >
                      > -----
                      > To unsubscribe send a message to: svg-developers-
                      > unsubscribe@yahoogroups.com
                      > -or-
                      > visit http://groups.yahoo.com/group/svg-developers and click "edit
                      > my membership"
                      > ----
                      > Yahoo! Groups Links
                      >
                      >
                      >
                    • T Rowley
                      ... Correction - gecko does have a mechanism for splitting style information into repaint/reflow, SVG just isn t using it yet. The fun of depreciated APIs
                      Message 10 of 23 , Feb 13, 2007
                      View Source
                      • 0 Attachment
                        On 2/13/07 3:37 PM, T Rowley wrote:
                        > Changing the style system to give detailed information is a larger scope
                        > item, though maybe something gecko would want to do as it could help
                        > other sorts of content.

                        Correction - gecko does have a mechanism for splitting style information
                        into repaint/reflow, SVG just isn't using it yet. The fun of
                        depreciated APIs floating around...

                        -tor
                      • T Rowley
                        ... This appears to be one of the SVG applications having a regression of the onload behavior. It is fixed in my local tree which contains the patch from
                        Message 11 of 23 , Feb 13, 2007
                        View Source
                        • 0 Attachment
                          On 2/13/07 2:58 PM, Guy Morton wrote:
                          > I wish I could report the same happy news. Unfortunately,
                          > GranParadiso (on MacOSX at least) fails to render *anything* for me.
                          > It even seems buggy in doing normalish browser things like resizing
                          > framesets (the toolbar has half disappeared under the window title in
                          > my testing).
                          >
                          > Have a look at http://bct2.webtrak-lochard.com/ and see.
                          >
                          > It's not generating anything in the error logs that gives any
                          > indication of what it might be taking exception to either.

                          This appears to be one of the SVG applications having a regression of
                          the "onload" behavior. It is fixed in my local tree which contains the
                          patch from bug 370210.

                          -tor
                        • T Rowley
                          ... In your case, this is probably largely due to the background image you re using and changes in how we implement internally. ... I don t see
                          Message 12 of 23 , Feb 13, 2007
                          View Source
                          • 0 Attachment
                            On 2/13/07 3:41 PM, Guy Morton wrote:
                            > Aha...well that's the problem for me. I resized the window and it
                            > appeared. Good to know you know what is causing it.
                            >
                            > And I'll add my "Wow" to Andre's...FF used to use 70% of my CPU
                            > playing this file. Now it's using about 7%.

                            In your case, this is probably largely due to the background image
                            you're using and changes in how we implement <svg:image> internally.

                            > There are some screen
                            > redrawing issues though still, which you'll see if you look at the
                            > link I sent (the planes are leaving translucent colour over the map).

                            I don't see redraw problems on linux here. If you could send me a
                            screenshot I can look into the problem a bit.

                            -tor
                          • TheMountainScene
                            It was fine until two days ago when I played with it a bit. I ll put the old one back up. Thanks!
                            Message 13 of 23 , Feb 13, 2007
                            View Source
                            • 0 Attachment
                              It was fine until two days ago when I played with it a bit. I'll put the
                              old one back up. Thanks!

                              T Rowley wrote:
                              >
                              > On 2/13/07 2:24 PM, TheMountainScene wrote:
                              > > In a FF test I've created, a mouseover of Canada 364 Kb, it has
                              > improved
                              > > from 6 seconds to 2 seconds. 2 Seconds is still much too slow compared
                              > > to the Adobe viewer, but it is an improvement. I submitted this as a
                              > bug
                              > > some time ago. For my purposes, it still falls short, and given the
                              > > snail's pace anything svg is moving, I see flash in my future. I
                              > keep my
                              > > fingers crossed, but my optimism wanes with each passing day.
                              >
                              > Assuming you're talking about the Canada map in bug 319990, the version
                              > you have online right now has a JS error. Correcting that, removing the
                              > alert, and moving in/out right near the border (which defeats our
                              > early-reject hit detection), the major hotspots in the profile all
                              > appear inside the cairo 2D graphics library:
                              >
                              > 57.20% checking if a point is inside the path
                              > 99.7% of this time is spent tesselating the path to trapezoids
                              >
                              > 14.83% painting paths
                              > 99% of this time is spent tesselating the path to trapezoids
                              >
                              > 14.51% getting screen path extent
                              > 99.9% of this time is spend tesselating the path to trapezoids
                              > The reason this appears in the profile is because the style change
                              > information Gecko gives us is a general "things changed" rather than
                              > detail which would let us realize that only the fill changed and the
                              > extent didn't need to be recalculated.
                              >
                              > It seems that if cairo improved their point-in-path functionality using
                              > one of the techniques that doesn't need a whole trapezoid list, things
                              > would be a fair bit faster.
                              >
                              > Changing the style system to give detailed information is a larger scope
                              > item, though maybe something gecko would want to do as it could help
                              > other sorts of content.
                              >
                              > It does seem as though this map is a bit of a torture case for cairo's
                              > tesselator - I've passed on the URL to the cairo developers.
                              >
                              > -tor
                              >
                              >
                            • TheMountainScene
                              The original is back up. Online, the version is still much faster, 2 vs 7 or 8 seconds. If it can get to Adobe performance, I d be happy as a clam.
                              Message 14 of 23 , Feb 13, 2007
                              View Source
                              • 0 Attachment
                                The original is back up. Online, the version is still much faster, 2 vs
                                7 or 8 seconds. If it can get to Adobe performance, I'd be happy as a clam.

                                T Rowley wrote:
                                >
                                > On 2/13/07 2:24 PM, TheMountainScene wrote:
                                > > In a FF test I've created, a mouseover of Canada 364 Kb, it has
                                > improved
                                > > from 6 seconds to 2 seconds. 2 Seconds is still much too slow compared
                                > > to the Adobe viewer, but it is an improvement. I submitted this as a
                                > bug
                                > > some time ago. For my purposes, it still falls short, and given the
                                > > snail's pace anything svg is moving, I see flash in my future. I
                                > keep my
                                > > fingers crossed, but my optimism wanes with each passing day.
                                >
                                > Assuming you're talking about the Canada map in bug 319990, the version
                                > you have online right now has a JS error. Correcting that, removing the
                                > alert, and moving in/out right near the border (which defeats our
                                > early-reject hit detection), the major hotspots in the profile all
                                > appear inside the cairo 2D graphics library:
                                >
                                > 57.20% checking if a point is inside the path
                                > 99.7% of this time is spent tesselating the path to trapezoids
                                >
                                > 14.83% painting paths
                                > 99% of this time is spent tesselating the path to trapezoids
                                >
                                > 14.51% getting screen path extent
                                > 99.9% of this time is spend tesselating the path to trapezoids
                                > The reason this appears in the profile is because the style change
                                > information Gecko gives us is a general "things changed" rather than
                                > detail which would let us realize that only the fill changed and the
                                > extent didn't need to be recalculated.
                                >
                                > It seems that if cairo improved their point-in-path functionality using
                                > one of the techniques that doesn't need a whole trapezoid list, things
                                > would be a fair bit faster.
                                >
                                > Changing the style system to give detailed information is a larger scope
                                > item, though maybe something gecko would want to do as it could help
                                > other sorts of content.
                                >
                                > It does seem as though this map is a bit of a torture case for cairo's
                                > tesselator - I've passed on the URL to the cairo developers.
                                >
                                > -tor
                                >
                                >
                              • ddailey
                                I ve got a few dozen examples that work fine in FF1.5, Opera 9 and IE/ASV3, but not in Gran Paradiso. The stuff that does work in Gran Paradiso, though, is
                                Message 15 of 23 , Feb 14, 2007
                                View Source
                                • 0 Attachment
                                  I've got a few dozen examples that work fine in FF1.5, Opera 9 and IE/ASV3, but not in Gran Paradiso.
                                  The stuff that does work in Gran Paradiso, though, is very much faster to load, and in some cases, to run script than what I'm accustomed to.
                                  It appears that if there is any SMIL whatever, then Gran Paradiso doesn't render the file. I think FF1.5 displays the static image.

                                  Is this the appropriate place to post examples?

                                  regards
                                  David

                                  [Non-text portions of this message have been removed]
                                Your message has been successfully submitted and would be delivered to recipients shortly.