Hey sorry for the delay - I ended up getting pretty sick over the
weekend and haven't been back into the office until today.

I was all setup to run your test on Friday under IE8 - Chrome - and
Firefox, but the interval on calling Cycle was changed from the
previous 3 seconds to 0 ms.  Opening the test in those 3 browsers
(each full-screen on one of 3 monitors) had my CPU usage hanging out
at around 60%+.  I didn't really want to tax the work machine that
badly over the weekend so I elected to not run the test.

Anyhow I am glad that our discussions helped you find some ways to
improve your library - it is a very impressive.

I definitely recommend using SVG as a default over CANVAS.  Though
both SVG and CANVAS require the browser to render the individual
pixels when they are displayed - CANVAS, being bitmap based, has to
store a value for what it is that it is going to draw on every point -
unlike SVG which uses a more mathematical vector approach.  In testing
- I found that very very very large CANVAS objects took significantly
more memory to render than SVG objects of the same size.

I also found that increasing the scale of an SVG object doesn't
necessarily increase the memory usage - if we are only talking about
rendering - when the SVG object falls outside of the viewpoint of the
browser itself (if the object was being drawn off-screen).

I am going to continue working on tweaking my own custom solution -
which overlays very large SVG objects over the map once - then alters
those objects directly whenever data changes, instead of deleting and
re-creating new ones.  Time to refresh my memory on how to embed those
VML script and namespace tags.

Thanks again for your help as well - it feels significantly better
knowing that somebody else out there has some real experience with
problems like this.   (I've felt so alone for awhile!  Haha)


On Dec 6, 12:05 pm, bratliff <[email protected]> wrote:
> On Dec 3, 6:07 pm, John Mick <[email protected]> wrote:
>
> > I'll be leaving for the weekend in a few hours - when I leave today
> > I'll clear out all my cache and make sure I am loading your SVG
> > version test and let that run over the weekend, so we'll have several
> > days of usage to look at on Monday.
>
> Did anything crash ?
>
> I added five copies of each poly to the test case.  I ran it overnight
> on a very old computer.  It was still running in the morning with no
> change in speed.
>
> I have made several improvements.  SVG is the default.  I might
> discard CANVAS.
>
> Your application is fairly unique.  It helped me isolate & fix some
> bugs.  It is not safe to simply rebuild dirty tiles.  It causes the
> zIndex order to change.  Adjacent tiles with different zIndex orders
> are apparent.  If a poly changes position, I have to do a full
> rebuild.  I believe it is working now.
>
> Your custom solution will provide the best performance.  Both
> PolyCluster & the API build tiles to expedite dragging, panning &
> zooming.  Since you are doing none, tiles are overkill for you.
> PolyCluster builds an internal cache to reduce the effort to rebuild
> tiles previously in view.  Again, because you quickly discard your
> polys, an internal cache is overkill for you.
>
> Thanks again for your help.

-- 
You received this message because you are subscribed to the Google Groups 
"Google Maps JavaScript API v3" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/google-maps-js-api-v3?hl=en.

Reply via email to