Dicing user-specified params (GSoC)

[Trevor Lovett]

In curves.cpp I've overridden CqSurface::Dice() as we discussed and I would like to begin the rework of the dicing of the user-specified params.

Where I will be writing new code is inside this loop:

std::vector<CqParameter*>::iterator iUP;

std::vector<CqParameter*>::iterator end = m_aUserParams.end();

for ( iUP = m_aUserParams.begin(); iUP != end ; iUP++ ) 

{ ...}

To summarize my understanding of the problem:

We can't use SetArgument (what exactly is an "argument" ? ) because further down the line that leads to a Dice call (on a primvar) which assumes bilinear interp instead of linear.  So, instead we need to do a pShader->FindArgument to get to the IqShaderData* and then, lerping begins along v: we compute the appropriate offset and then do a IqShaderData::SetValue(foo, offset) call.  Also, a detail: I guess we need to do a big switch (a la NaturalDice) on *iUP to tell what type of parameter it is, then do (for example): 

     CqParameterTyped<TqFloat, TqFloat>* pTParam = static_cast<CqParameterTyped<TqFloat, TqFloat>*>(*iUP);

before finally calling the appropriate SetValue function.

...

Finally, to my question:  how should I use FindArgument?  I see that it wants a string as the "argument" name.  Where do I get this value? 

A wild guess: (*iUP)->strName() ?

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Aqsis-development mailing list
Aqsis-development@...
https://lists.sourceforge.net/lists/listinfo/aqsis-development

[Chris Foster-5]

On Wed, Aug 12, 2009 at 7:57 AM, Trevor Lovett<trevlovett@...> wrote:

> In curves.cpp I've overridden CqSurface::Dice() as we discussed and I would
> like to begin the rework of the dicing of the user-specified params.
>
> Where I will be writing new code is inside this loop:
> std::vector<CqParameter*>::iterator iUP;
> std::vector<CqParameter*>::iterator end = m_aUserParams.end();
> for ( iUP = m_aUserParams.begin(); iUP != end ; iUP++ )
> { ...}
>
> To summarize my understanding of the problem:
> We can't use SetArgument (what exactly is an "argument" ? )

An "argument" is a formal parameter to a shader function (in the RISpec I
think they call this an "instance variable" for the shader).  That is, in the
shader

surface my_surf(color col = (1,0,0)) { /*...*/ }

"col" is an "argument".

> because further
> down the line that leads to a Dice call (on a primvar) which assumes
> bilinear interp instead of linear.

Correct.

> So, instead we need to do
> a pShader->FindArgument to get to the IqShaderData* and then, lerping begins
> along v: we compute the appropriate offset and then do a
> IqShaderData::SetValue(foo, offset) call.

Actually, for efficiency I'd try to avoid SetValue and instead set values
directly through a pointer.  SetValue involves calling at least one virtual
function call (yuck, yes this interface should go away in future!).  That is,
call GetValuePtr to get a raw pointer to the data instead.

>  Also, a detail: I guess we need
> to do a big switch (a la NaturalDice) on *iUP to tell what type of parameter
> it is, then do (for example):
>      CqParameterTyped<TqFloat, TqFloat>* pTParam =
> static_cast<CqParameterTyped<TqFloat, TqFloat>*>(*iUP);
> before finally calling the appropriate SetValue function.
> ...

Yes, you're going to need a big switch, possibly like in
CqSurface::NaturalDice.  Have a read of that function, and you'll see that I
factored all the type-specific stuff into a function "surfaceNaturalDice".
Depending on the details, that's probably the right thing to do.

> Finally, to my question:  how should I use FindArgument?  I see that it
> wants a string as the "argument" name.  Where do I get this value?
> A wild guess: (*iUP)->strName() ?

Correct!  Just look inside CqShaderVM::SetArgument() and you'll see the
analogous thing.

Cheers,
~Chris.

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Aqsis-development mailing list
Aqsis-development@...
https://lists.sourceforge.net/lists/listinfo/aqsis-development