Re: More changes in path drawing

Hi Robert,

On Tue, Feb 15, 2011 at 01:54, Robert Park <rbpark exolucere ca> wrote:
> On Mon, Feb 14, 2011 at 4:49 PM, Jiří Techet <techet gmail com> wrote:
>> ChamplainLocation - interface for objects with latitude and longitude
>> ChamplainCoordinate - simple gobject implementing ChamplainLocation
>> ChamplainMarker - now implements ChamplainLocation too
>> ChamplainPathLayer - insert objects implementing ChamplainLocation to
>> draw the path
>> ChamplainMarkerLayer - now doesn't contatin the path-drawing methods
>> Please test and let me know if the performance got better.
> Ok, We're now down to 9.36 seconds to load the big benchmark file with
> path visibility turned off (this confused me at first because before,
> visibility defaulted to off, and I had to specifically turn visibility
> on, but now it defaults to on and I have to specifically turn it off
> in order to see the speed benefit. It's still quite slow (several
> minutes, didn't have time to let it run to completion) with path
> visibility turned on.

Paths should be drawn on idle now so it shouldn't be necessary to turn
off the path display before marker addition. As for the speed, we can
later try if a function taking a list of coordinates speeds up things
but as it is a minor change and there's no dramatic regression now, I
would concentrate on making a release as soon as possible. I expect
the performance will be slightly worse than in 0.8 -
ChamplainCoordinate is a gobject now so there is some overhead for
creating the object itself.

> This is a good improvement, though! Thanks so much!
> I'm having a problem with MarkerLayers now though, I can't seem to get
> them to display. I've tried calling animate_in() and show() and they
> just don't appear. Did I miss something somewhere?

This is my first interface with gobject so it's possible I did
something wrong. What you say here

can be interpreted in other way too - the methods don't work for
subclasses of CamplainMarker. I implemented the interface only for
ChamplainMarker assuming this was enough but maybe I have to do
something more in the subclasses too - I have to check. On the other
hand it works in C. One more strange thing is that the documentation
for the interface doesn't look the same way as documentation of
interfaces of other packages, e.g. ClutterContainer. This would
suggest that the annotations are not alright, but I checked at least
three times and I do the very same thing as clutter does for interface


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]