[sldev] Implementing extensions in the Open Source client only.

[Argent Stonecutter]

Sorry about the last digest being mostly my responses, they were  
delayed an "crossed in the post".

Getting back to the Open Source client... :)

> I got a new Feature Request about getting feedback of the touched  
> coordinates on a side of a prim. This would enable to save prims on  
> HUDs and other control panel-like objects.
>
> (Think "imagemap" in html.. basically.)
>
> https://wiki.secondlife.com/wiki/Feature_Request_Touch_Coordinates
>
> Please discuss there and feel free to add to the pros and cons.

Can you think of a way that this could be implemented using the open  
source client code only?

I don't see a way to do it using the ClickAction message, perhaps the  
client could detect clicks on a special tagged texture (shared common  
UUID, like the invisiprim UUIDs) and send a formatted message to a  
well known channel?

I was thinking of special tagged textures like that today for  
implementing mirrors, using a variant on the freeze-frame snapshot  
mechanism.

When a texture with a specific UUID was displayed on a flat face (to  
simplify things, you wouldn't allow curved mirrors), each frame the  
client would run a rendering pass from the mirrored viewpoint clipped  
to the mirror's bounds, and render the resulting texture in place of  
the mirror. This should take significantly less time than a full  
frame unless the mirror is almost filling the frame... in which case  
most of the frame would be masked. The case of multiple mirrors could  
be ignored... only the closest mirror to the avatar would be rendered  
as a mirror, the rest would show the underlying texture.

This could be done without any changes to the sim-side software.

What are the downsides of this approach?

1. Getting Linden Labs to pick up the code after it's done. If it's  
cool enough that shouldn't be a problem.

2... ?