Thanks Rob
Very useful info, noticed the draw primitive change as it effected my
hacking, but since still lots to learn not noticed much else! I know
it's a lot to ask but have you considered if this will have any impact
on your book is there a place where users can add any notes on changes
if we find them while reading, like a wiki page or is there one already?
Cheers
;j
On 6 Oct 2010, at 21:02, Rob Bateman wrote:
Hey guys
so, hopefully this mail will aid comprehension and reduce
unnecessary suffering, always a good thing ;)
As some of you already know, Away3D 3.6.0 was launched some time
last week with a few unavoidable refactors that will regrettably
have an effect on any projects currently in production. In fact, at
present my advice would be to not upgrade to 3.6.0 for any projects
currently in production. If you are desperate for some of the
optimisations but still don't want the hassle of refactoring (man,
you're so lazy!), then we have a nice surprise for you - many of the
upgrades are available in a recently tagged 3.5.2 version that you
can find in the googlecode svn here:
http://away3d.googlecode.com/svn/tags/3.5.2%20lib/
this obviously won't be as fast as the latest version, but should at
least give you some speed and memory enhancements.
As for the peeps still using Flash 9, this has also recently been
upgraded in a similar style to 2.5.2 - again something you can find
tagged here:
http://away3d.googlecode.com/svn/tags/2.5.2%20lib/
On to the main event, the 3.6.0 release. This has had a few fairly
fundamental changes to its structure, a breakdown of which follows.
to start with:
Number3D is replaced by the native Vector3D class.
MatrixAway3D is replaced by the native Matrix3D class.
while these two changes look fairly innocuous, they have far-
reaching implications to existing project code - basically any 3d
number in the library is now represented by a Vector3D instance, and
any 3d transform matrix is now reprsented by a Matrix3D instance.
Another big difference is the way these classes operate on each
other. Previously, Away3D used methods on Number3D and MatrixAway3D
that avoided the creation of new class instances, to save memory.
With the use of native classes this issue is no longer a problem,
but it does mean that something like:
myDifferenceVector.sub(myObject1.position, myObject2.position);
now looks like:
myDifferenceVector = myObject1.position.subtract(myObject2.position);
Another API change involves the location of session classes -
previously these were lumped together with render classes and called
SpriteRenderSession, BitmapRenderSession. Now these have been moved
to their own package inside away3d.core.session and have been
renamed SpriteSession, BitmapSession. Could we possibly be planning
more session types?
There have already been a few confused posts on the list about
sprites - apologies for having to wait so long for an explaination
on this. Basically, sprites are now defined as an extension of an
Element class (the same as Face and Segment), and have a root class
called Sprite3D. This accepts a material rather than a bitmap, and
an instance of Sprite3D is added to the scene using an addSprite()
method. users of the deprecated Sprite2D class an easily simulate
all the previous properties by wrapping a Sprite3D instance inside
an ObjectContainer3D instance. This may seem like more hassle than
its worth, but the main reason for the change has once again been
about speed and unification.
Similarly to sprites, all light classes now inherit from their own
AbstractLight class rather than Object3D, and as a result are much
more self explanatory to use. They are now added to a scene using
the method addLight().
Finally, for those of you poking a bit deeper into the core of the
library, you'll notice that all draw primitives have been removed -
this is something we've wanted to do for a while now but its been
quite a task! essentially any draw information is now stored inside
number vectors in the renderer. Triangle caching all still works as
normal, but without hundreds of objects being passed around every
frame, memory and framerates should benefit.
I'm sure there are other small alterations i've missed in this
initial post, so please let me know if there is something else
you've noticed that is different to the previous release!
atb
Rob
--
Rob Bateman
Flash Development & Consultancy
[email protected]
www.infiniteturtles.co.uk
www.away3d.com
