On Mon, Oct 02, 2006 at 11:07:26AM +0200, Udo Giacomozzi wrote: > >> The renderer could instruct the character definition to remove the > >> original RGB(A) buffer, but I don't like this at all. > > It would be a solution, but I'm not very convinced about it. For > example, a renderer could convert the bitmap to a 16 bpp or grayscale > buffer if that makes sense. However, when the bitmap is not used early > (or not used at all) the full RGB(A) buffer is kept in memory.. > > I'm just thinking about some map movies with images of about some > megapixels we used. With such movies memory is an issue.
I'm really thinking about just forget the render/parser split all togheter. After all registering a renderer before parsing shouldn't be such a big issue... we'd just need to clearly document this need. > s> Initially this was a threading issue, as the opengl backend would use > s> threads and those were messing the the loading/playing threads layout. > > What about semaphores / spinlocks / critical sections (whatever > they're called in C++)? As I said, that seems to not be a problem anymore. --strk; _______________________________________________ Gnash-dev mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnash-dev
