On Nov 26, 10:30 pm, "Michael Geary" <[EMAIL PROTECTED]> wrote:
Thanks Mike,
> GPolygon goes through multiple function calls for every point. You have to
> create a GLatLng for each point, ang GPolygon converts each lat/lng to a
> pixel coordinate by going through quite a bit of general purpose code.
It baffles me why people use an array of [new GLatLng(Lat,Lon)]
objects instead of an array of [{x:Lon,y:Lat}] objects to build a
GPoly object. The speed advantage of GPoly.fromEncoded() vanishes
with the later approach.
> PolyGonzo uses a tight inner loop for the points within a single polygon,
> with separate inner loops for IE (generating VML directly) and other
> browsers (using Canvas). It makes no JavaScript function calls in these
> inner loops, and precalculates as much of the coordinate conversion as
> possible outside the inner loop. This conversion is hard coded to use
> Mercator at the moment.
>
> In addition, all of the variables used in the inner loop are local
> variables. There's an eachShape() function in polygonzo.js that has a
> surprisingly long list of function parameters. These parameters are there so
> that the references to them are truly local variables. This saves on name
> lookups in the inner loop.
>
> Here's the innermost loop for IE:
>
> for( var iCoord = -1, coord; coord = coords[++iCoord]; ) {
> vml[iVml++] = round( coord[0] * 10 );
> vml[iVml++] = ',';
> vml[iVml++] = round( coord[1] * 10 );
> vml[iVml++] = ' l ';
> }
I might have joined elements in the outer loop but concatenated
elements in the inner loop. I am surprised four elements incur much
penalty.
Concatenation degrades for long strings exceeding their allocated
space.
str+=round( coord[0] * 10 );
str+=',';
str+=round( coord[1] * 10 );
str+=' l ';
is slow but
str[i++]=round( coord[0] * 10 )+','+round( coord[1] * 10 )+' l ';
ought to be fast.
The conversion from float to string is done automatically.
Is your own "round" function (interpreted) more efficient than the
built-in "Math.round" function (compiled) ?
> The use of "vml[iVml++] = ...." in the IE version is itself one of the
> optimizations. It's quite a bit faster than the "vml.push(...);" that I
> would prefer to use.
Interesting. I like to do the same thing using the loop counter for
an index but it is simply dumb luck.
> > For my own JS enlightenment, I am not sure I understand the
> > purpose of the unnamed / anonymous "()" function. Is it to
> > keep clutter out of the global namespace ? I believe Google
> > is using a similar trick with their classic loader. Do you
> > anticipate a conflict ?
>
> Yes, that's precisely what it's for - to create a local namespace. The
> particular form that I used in polymap.js and testmap.js is popular in
> jQuery code:
I have been looking for a way to avoid namespace pollution. I was
afraid to redefine the "()" function. Google's "main.js" already uses
it.
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Google Maps API" 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-API?hl=en
-~----------~----~----~----~------~----~------~--~---
