Alembic 1.1.2

[Barnaby Robson]

Hey Guys,

We were thinking of releasing 1.1.2 with the remaining straggling pieces
that didn't make it into 1.1 or 1.1.1.

Things going into the release would be the recent entries in
jihunyu-alembic11 plus the material assignment bindings in
rsgalloway-dev plus any maya plugin changes you have already
integrated. And maybe the Windows patch recently contributed ?

We were hoping for a target of this week. Does that seem ok with you ?
We can do the integration into trunk if you are busy.

Thanks !

barnaby.

[Lucas Miller]

Can we wait till next week?

I've got a child bounds change to the O side that I still want to make.

I believe Jihun's change here: http://code.google.com/r/jihunyu-alembic11/source/detail?r=ac84dff38689df4c0307d77a72f2b11af0c26bf5

is superseded by: http://code.google.com/r/millerlucas-dev/source/detail?r=ad1c43b5a06d8c4e516c6854528b2f0bd5cf5fb7

Also, can I get some clarification on why:

http://code.google.com/r/jihunyu-alembic11/source/detail?r=324d10a892e3bb665b73a3fdab47422c10d20026

is necessary?

Lucas

[Barnaby Robson]

Hi Lucas,

Yes, that sounds reasonable to wait. Maybe you should do the final integration then ?

I talked with Jihun about the changes and he agreed that your previous change looks good, thanks.

Regarding the last question; from Jihun:

".. the reason is that #define for const strings does not work outside "Alembic::AbcMaterial" namespace (i.e. Alembic::AbcMaterial::MATERIAL_PROPNAME), and we need to use it in PyAlembic."

barnaby.

[Lucas Miller]

".. the reason is that #define for const strings does not work outside "Alembic::AbcMaterial" namespace (i.e. Alembic::AbcMaterial::MATERIAL_PROPNAME), and we need to use it in PyAlembic."

On which compiler are you seeing this issue?

I didn't seem to have a problem in PyMaterialAssignment.cpp doing:

     def( "addMaterialAssignment", 

         addMaterialAssignmentToObject,

         ( arg( "iObject" ), 

           arg( "iValue" ), 

           arg( "iPropName" ) = MATERIALASSIGN_PROPNAME )

Now if you instead you tried to do:

arg( "iPropName" ) = Alembic::AbcMaterial::MATERIALASSIGN_PROPNAME

There would be a problem.

Lucas

[Ji Hun Yu]

Aha, That's strange. I don't see the problem either.
I believe that the error report was originally from Ryan and he had hard time making it work.
I think we are good to go with defined strings then.

Thanks,
 
- Jihun

---

From: alemb...@googlegroups.com [alemb...@googlegroups.com] on behalf of Lucas Miller [miller...@gmail.com]
Sent: Tuesday, October 02, 2012 6:42 PM
To: alemb...@googlegroups.com
Subject: Re: Alembic 1.1.2

[Ryan Galloway]

I could have sworn it didn't work for me before. I can try to reproduce the issue tomorrow if it's informative at all.

Ryan Galloway
Global Pipeline Engineer

---

From: alemb...@googlegroups.com [alemb...@googlegroups.com] on behalf of Ji Hun Yu [jih...@ilm.com]
Sent: Tuesday, October 02, 2012 6:54 PM
To: alemb...@googlegroups.com
Subject: RE: Alembic 1.1.2

[Lucas Miller]

Here is the child bounds change I mentioned I was going to make:

http://code.google.com/r/millerlucas-dev/source/detail?r=f143ab3b6da879a727032c7e2a5bc1cae8444048

If this change isn't too controversial, I'll make a similar change to the python bindings.

From my commit log, here is what I did and why I think it's necessary:

Remove child bounds from being set and created as part of the samples on the
OSchemas. Instead give access to the OBox3dProperty child bounds directly.

In many cases where child bounds is created, the child objects are sampled
more frequently than the schema itself.
(static xform, animated child polymeshes for example).

Child bounds is optional data that isn't influenced directly by the data on
the owning OSchema.

I'm debating whether this should be done for the ISchema as well.

They already have access to IBox3dProperty via getChildBounds and getting the I*Schema::Sample will return the correct value for the child bounds, even if it's different from the I*Schema sampling. (because of the SampleSelector)

I'm just worried about it being a little confusing, since you would have to use getChildBounds anyway to figure out that the time sampling is different.

Lucas