Hi all, mainly cedric that is the author of the last patches of preload:
We're missing some cases to call EVAS_CALLBACK_IMAGE_PRELOADED, it
should be called in any case after evas_object_image_preload() is
called, if it fail anyhow or so, otherwise people will never know how
I suffered this with software that was chaining load of images. In my
case, the chain was broken since I'm not getting the callback in some
cases (still unknown which are the case roots, otherwise I'd fix it).
The initial version of preload had similar bugs as well, that I had to
fix... now I need your help as I can't dig into it :-)
NOTE: You don't need to callback just if you are explicitly cancelling
the preload (calling it with second parameter=0).
Last but not least, the way preload is done right now will break badly
if loader does not support being used from different threads. This is
the case with XPM loader and I guess SVG, but may be the case with
others. To solve this, either fix the loaders to queue requests, or
make infrastructure to queue loaders that do not support that, so you
never start different threads with that loader and do not use it from
worker and main thread.
Gustavo Sverzut Barbieri
http://profusion.mobi embedded systems
Mobile: +55 (19) 9225-2202