On 29-07-09 12:38, Emmanuele Bassi wrote:
On Wed, 2009-07-29 at 12:28 +0200, Koen Kooi wrote:
On 29-07-09 11:10, Emmanuele Bassi wrote:
On Wed, 2009-07-29 at 10:12 +0200, Koen Kooi wrote:
Today I want to try clutter with GLES but failed to compiled for some
missing variables such as /model_view, projection_stack…/
Looks a lot like this bug that's still open:
http://bugzilla.openedhand.com/show_bug.cgi?id=1698
I really wish that clutter people who break such stuff would clean up
after themselves....
and I'd really wish for GL|ES users to submit patches since we explained
multiple times that we cannot currently test GL|ES backends.
let me re-state this as clearly as I can: compiling *is* *not* *the*
*problem*: there have been changes in how Clutter and COGL work, and GL|
ES backends have *not* been ported to them because we cannot *test* it.
And people aren't going to test it since it doesn't compile. Also, is
downloading and installing the powervr simulator or an opensource
gl->gles bridge that hard?
the PowerVR simulator has its own idiosyncrasies that might make stuff
compile, but break subtly; this is acquired knowledge from the 0.8
development cycle. the only way of actually verifying that the GL|ES
backends work with real implementations is to have real implementations
run the code. which we don't have.
OK, understood. I still maintain that people aren't going to test stuff
that doesn't compile. I can get people to debug runtime problems, but I
can't get them to fix the build first.
The cynical side of me suggests that intel is blocking GLES work so
clutter only works on x86, but that's just me being cynical.
that would be just you being cynical and wrong. we helped, and still
help everyone that wishes to use Clutter on GL|ES hardware. if we wanted
to mess people up we could have just yanked out the whole GL|ES backends
from the start.
You should yank then from the releases, but keep them in git. That way
it's clear that GLES is broken and basically unsupported, but leaves the
door open for someone to fix it.
regards,
Koen
--
To unsubscribe send a mail to [email protected]
