[Soc-2013-dev] Weekly Report #'s 11-13 Depsgraph Refactor

Joshua Leung aligorith at gmail.com
Fri Sep 13 16:31:49 CEST 2013


Hi all,

== Week 11 ==
Not much progress to report, as this was quite a busy week at uni - the
annual departmental postgraduate students conference was on in second half
of that week, and I needed to give a presentation there.

== Week 12 ==
I managed to fit in a bit of work on bits and pieces of depsgraph work, but
a heavy writing+coding workload for uni work took up almost all my time
that week again. Also, started discussing overall design with Sergey.

== Week 13 - This Week ==
Starting a few days ago, I now finally have time to seriously work on
Depsgraph stuff again (at least until the end of GSoC period, and then some
;)

I started by taking stock of the tasks which still need to be done in order
to start integrating the new Depsgraph module with the main Blender
codebase. In particular, I've divided the work into several categories:
1) Before Integration Todos - Minimal Depsgraph Functionality/Core Code in
Place. To keep things manageable, this means that some of the more
problematic areas in Blender will be left alone for now (see Part 3) so
that we can concentrate on ensuring that the core parts are put in place
and holding well.
2) Integration Todos - Things like resolving compiler errors/typos,
adapting evaluation functions (i.e. those called via
object_handle_update(), but also the ModifierTypeInfo->updateDepsgraph()
callbacks), replacing existing evaluation system entrypoints with our new
system
3) Post Integration Todos - These are things we won't get around to before
the end of GSoC. Coding final solutions for the issues in this category
(e.g. proxies, instancing, metaballs) will have to wait a bit - there are
various bits and pieces we can put together in the new system to eventually
resolve these issues, but dealing with each one is itself quite a challenge.

Accomplishments so far:
- 1/3rd of the before integration todos complete. Of the remaining parts,
most should be quick to blast through, while one or two will require a bit
more work. Namely, fleshing out the details of evaluation contexts and
feeding nodes to the scheduler - will be discussed with Sergey some more in
coming days.
- I'm currently in the process of cleaning up compiler errors in the code
(2.5/5 source files done atm), and collecting a list of stubs for
evaluation callbacks which need to be implemented.

== Next Week ==
I'll try to get things as close as possible to a basic "running" state
(albeit, only with a very limited "safe" feature set), which we can start
to seriously flesh-out post GSoC.


Regards,
Joshua
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2013-dev/attachments/20130914/4a364a76/attachment.htm 


More information about the Soc-2013-dev mailing list