agg changes again...

[Paul Spencer]

My opinion (noting that I don't have a vote) is that it would be fine  
to release the first RC with this change, another beta would not  
accomplish any more than delay release.  Those that are using/testing  
AGG will do so with the first RC and report issues that can be fixed  
in the next RC or the release.  You won't find out some issues until  
after 5.0 is released and a wider audience starts adopting it, but  
that is okay too (that's what 5.0.x is for).

I'd like to see this change in 5.0

Paul


On 1-Sep-07, at 9:10 AM, thomas bonfort wrote:

> hi all....
>
> I'vee finally managed to get AGG working directly on a gdImg,
> completely ignoring it's alpha channel, therefore not having to switch
> the alpha channel to/from gd/agg before and after rendering.
>
> this means that it will no longer be necessary to call msAlphaAGG2GD
> and msAlphaGD2AGG which is a good thing as they added some overhead
> and were very hacky (except of course when using RGBA imagemode, as in
> this case the alpha values in the image /are/ meaningfull)
>
> I see two paths from this:
> 1/ I commit this for 5.0. we get a bit faster rendering especialy for
> maps with lots of layers, but this warrants at least another beta
>
> 2/ we wait with the current solution and incorporate these changes  
> in 5.2
>
>
> I'll start with a +1 for solution no2, but people with performance
> issues (Zak notably) may have another opinion on this.
>
> thomas

+-----------------------------------------------------------------+
|Paul Spencer                          pspencer at dmsolutions.ca    |
+-----------------------------------------------------------------+
|Chief Technology Officer                                         |
|DM Solutions Group Inc                http://www.dmsolutions.ca/ |
+-----------------------------------------------------------------+