Closed pagameba closed 10 years ago
sorry about that, hopefully this is right now.
Thanks, no problem. I'm checking out the items from your notes. Definitely seems like there are some performance issues with some of the examples, especially when zooming/panning as you note. You're right, the conflict data example is really slow even in the released version. I swear it didn't used to be that way. I suspect your thinking is right on the hovering - it looks like elements are being given a new fill property with a different id that doesn't exist. I'm guessing this has to do with needing to recreate/replace the items in the defs element on redraw/update style calls.
The sochistats example is now defunct since Kimono Labs took that API away (wish I would have saved the sample response somewhere). It also looks like in the weather example the weather API I was using has changed as well, so I'll do some updating there to get it working again. The cross-origin issue with the geohashart example is probably due to changes in 0.8 related to tile layers - I'll look into updating that one as well.
Good to know it wasn't just my changes :)
Looks like most of the performance related issues are due to drop shadows. I know these tend to slow down SVG rendering performance, but they seem to be particularly slow here. Not sure if it's drop shadow performance in general or the way in which drop shadows are being applied here. I'll just turn off drop shadows in most of the examples until I can investigate further.
(Now rebased on 0.8dev) These commits add changes required to get all the examples running (as much as I could). Some notes:
geohashart doesn't run because of a cross origin error, not sure why? sochistats doesn't run because the data doesn't load, some kind of API error weather doesn't seem to do much There are some outstanding issues that I have noticed
in many of the examples, hovering over markers causes them to disappear or change style (permanently becoming black). I wasn't sure what triggered the hover behaviour, I suspect that the problem has to do with how patterns, fills etc are added to and removed from the _defs object in the renderer. some of the examples seem to run really slowly when interacting with the map (pan, zoom etc). I'm not sure if this is a regression or not - I tried to run the conflictdata example from the hosted version and it crashes chrome (Aw, snap). The 0.8 version does run, but zooming or panning is very slow and sometimes crashes the tab.