NAME¶

SoMField - The SoMField class is the base class for fields which can contain multiple values.

SYNOPSIS¶

#include <Inventor/fields/SoMField.h>

Inherits SoField.

Inherited by SoMFBool, SoMFColor, SoMFColorRGBA, SoMFDouble, SoMFEngine, SoMFEnum, SoMFFloat, SoMFInt32, SoMFMatrix, SoMFName, SoMFNode, SoMFPath, SoMFPlane, SoMFRotation, SoMFShort, SoMFString, SoMFTime, SoMFUInt32, SoMFUShort, SoMFVec2b, SoMFVec2d, SoMFVec2f, SoMFVec2i32, SoMFVec2s, SoMFVec3b, SoMFVec3d, SoMFVec3f, SoMFVec3i32, SoMFVec3s, SoMFVec4b, SoMFVec4d, SoMFVec4f, SoMFVec4i32, SoMFVec4s, SoMFVec4ub, SoMFVec4ui32, and SoMFVec4us.

Detailed Description¶

The SoMField class is the base class for fields which can contain multiple values.

All field types which may contain more than one member value inherits this class. SoMField is an abstract class.

Use setValue(), setValues() or set1Value() to set the values of fields inheriting SoMField, and use getValues() or the index operator [] to read values. Example code:

SoText2 * textnode = new SoText2;
textnode->ref();
// Setting multi-field values. /////////////////////////////
// Set the first value of the SoMFString field of the SoText2 node.
// The field array will be truncated to only contain this single value.
textnode->string.setValue("Morten");
// Full contents of the SoMFString is: [ "Morten" ]
// The setValue() method and the = operator is interchangeable,
// so this code line does the same as the previous line.
textnode->string = "Peder";
// Full contents of the SoMFString is now: [ "Peder" ]
// Set the value at index 2. If the field value array contained
// less than 3 elements before this call, first expand it to contain
// 3 elements.
textnode->string.set1Value(2, "Lars");
// Full contents of the SoMFString is: [ "Peder", <undefined>, "Lars" ]
// This sets 3 values of the array, starting at index 5. If the
// array container had less than 8 elements before the setValues()
// call, the array will first be expanded.
SbString s[3] = { "Eriksen", "Blekken", "Aas" };
textnode->string.setValues(5, sizeof(s) / sizeof(s[0]), s);
// Full contents of the SoMFString is now:
//     [ "Peder", <undefined>, "Lars", <undefined>, <undefined>,
//       "Eriksen", "Blekken", "Aas" ]
// Note also that the setValues() call will *not* truncate a field
// container if you use it to change a subset at the start:
SbString n[4] = { "Dixon", "Adams", "Bould", "Winterburn" };
textnode->string.setValues(0, sizeof(n) / sizeof(n[0]), n);
// Full contents of the SoMFString is now:
//     [ "Dixon", "Adams", "Bould", "Winterburn", <undefined>,
//       "Eriksen", "Blekken", "Aas" ]
// Inspecting multi-field values. //////////////////////////
// This will read the second element (counting from zero) if the
// multivalue field and place it in "val".
SbString val = textnode->string[2];
// Gives us a pointer to the array which the multiple-value field
// is using to store the values. Note that the return value is
// "const", so you can only read from the array, not write to
// it.
const SbString * vals = textnode->string.getValues(0);
// Modifying multi-field values. ///////////////////////////
// You can of course modify multifield-values by using the set-
// and get-methods shown above, but when you're working with
// big sets of data, this will be ineffective. Then use this
// technique instead:
SbString * modvals = textnode->string.startEditing();
// ... lots of modifications to the "modvals" array here ...
// Calling the finishEditing() method is necessary for the
// scene graph to be updated (and re-rendered).
textnode->string.finishEditing();

The reason it is more effective to wrap many modifications within startEditing() / finishEditing() is because we avoid the stream of notification messages which would otherwise be sent for each and every modification done. Instead there will just be a single notification about the changes, triggered by the finishEditing() call.

The correct manner in which to pre-allocate a specific number of field values in one chunk is to use the SoMField::setNum() method, for instance in advance of using the startEditing() and finishEditing() methods. The field values will be uninitialized after expanding the field with the setNum() call.

Be aware that your application code must be careful to not do silly things during a setValues()-triggered notification. If you have code that looks for instance like this:

// update set of coordinate indices at the start of e.g.
// an SoIndexedFaceSet's coordIndex field..
ifs->coordIndex.setValues(0, runner->numIndices, runner->indices);
// ..then truncate to make sure it's the correct size.
ifs->coordIndex.setNum(runner->numIndices);

As setValues() might leave some elements at the end of the array that typically can be invalid indices after the first statement is executed, something can go wrong during notification if you have application code monitoring changes, and the application code then for instance triggers an action or something that tries to use the coordIndex field before it's updated to it's correct size with the setNum() call.

(Notification can in this case, as always, be temporarily disabled to be on the safe side:

somefield.enableNotify(FALSE);
somefield.setValues(...);
somefield.setNum(...);
somefield.enableNotify(TRUE);
somefield.touch();

This will guarantee that the setValues() and setNum() pair will be executed as an atomic operation.)

When nodes, engines or other types of field containers are written to file, their multiple-value fields are written to file in this format:

containerclass {


  fieldname [ value0, value1, value2, ...]


  ...
}

..like this, for instance, a Coordinate3 node providing 6 vertex coordinates in the form of SbVec3f values in its 'point' field for e.g. a faceset, lineset or pointset:

Coordinate3 {


   point [


      -1 1 0, -1 -1 0, 1 -1 0,


      0 2 -1, -2 0 -1, 0 -2 -1,


   ]
}

Some fields support application data sharing through a setValuesPointer() method. setValuesPointer() makes it possible to set the data pointer directly in the field. Normally (when using setValues()), Coin makes a copy of your data, so this method can be very useful if your application needs the data internally and you're just using Coin for the visualization. Example code:

myapp->calculateCoordinates(SOME_LARGE_VALUE);
SbVec3f * mycoords = myapp->getCoordinates();
SoCoordinate3 * mynode = myapp->getCoordinateNode();
mynode->point.setValuesPointer(SOME_LARGE_VALUE, mycoords);

Be aware that your field should be a read-only field when you set the data like this. If you write to the field, the values in your application array will be overwritten. If you append values to the field, a new array will be allocated, and the data will be copied into it before appending the new values. The array pointer will then be discarded.

Also note that whenever you change some value(s) in the array, you must manually notify Coin about this by calling SoField::touch(). For our example:

SbVec3f * mycoords = myapp->getCoordinate();
myapp->updateCoordinate(mycoords);
SoCoordinate3 * mynode = myapp->getCoordinateNode();
mynode->point.touch(); // this will notify Coin that field has changed

You can use SoMField::enableDeleteValues() to make Coin delete the array for you when the field is destructed or the array pointer is discarded because it isn't needed anymore (e.g. when the array size is changed). The array will be deleted using the C++ delete[] operator, so if you use it, your array must be allocated using the C++ new[] operator.

SoMField::enableDeleteValues() is supported only to be compatible with later versions of TGS Inventor and we don't recommend using it. It can have undefined results on the Microsoft Windows platform. Allocating memory in the application and destructing it in a DLL can be a bad thing, causing mysterious crashes, if you're not very careful about how your application and DLLs are linked to the underlying C library.

See also

Constructor & Destructor Documentation¶

Member Function Documentation¶

Member Data Documentation¶

Author¶

Generated automatically by Doxygen for Coin from the source code.