
OpenGL::Array - Perl Array handling and conversion between Perl arrays and C array pointers.

use OpenGL qw(GL_FLOAT);
my $array = OpenGL::Array->new(4, GL_FLOAT);
my $c_ptr = $array->ptr(); # can be passed to OpenGL _c based functions
$array->calc('col,27,+');
my @val = $array->retrieve(0, 4);

OpenGL::Array (OGA) objects provide Perl Array handling and conversion between Perl arrays and C array pointers.
Due to the difference between how Perl and C handle pointers, all Perl OpenGL (POGL) APIs that require pointers are suffixed with _c. OGAs provide a means to convert Perl arrays into C pointers that can be passed into these APIs.
Many POGL _c APIs also have a _s version to support SDL's packed string APIs; OGA provides APIs to convert between C arrays and packed strings.
POGL also provides many _p APIs that accept native Perl arrays, or in some cases OGAs directly. In the case of VBOs, OGAs may be bound to GPU buffers, automatically switching buffers at render time.
Note: Since OGAs are stored as typed C arrays, there is no conversion/copy/casting when passing them to POGL APIs, resulting in significant performance improvements over other non-compiled bindings (SDL, PyOpenGL, etc).

newmy $array = OpenGL::Array->new($count,@types);
Creates an empty array object of $count rows made up data types @types.
new_listmy $array = OpenGL::Array->new_list($type,@data);
Creates and populates a uniform array object made up @data of type $type.
new_pointermy $array = OpenGL::Array->new_pointer($type,ptr,$elements);
Creates an array object wrapper around a C pointer ptr of type $type and array length $elements. Caches C pointer directly; does not copy data.
Note: because OpenGL::Arrays store to direct memory addresses, it is possible to assign to the array the pointer was obtained from and the results will be available in the array created by new_pointer - and vice versa (because they are viewing portions of the same memory).
new_scalar my $str = pack 'C*', 1 .. 255;
my $array = OpenGL::Array->new_scalar(GL_UNSIGNED_BYTE, $str, length($str));
Creates an array object from a perl scalar.
new_from_pointer my $array1 = OpenGL::Array->new_list(GL_UNSIGNED_BYTE, 1..9);
my $array2 = OpenGL::Array->new_from_pointer($array1->ptr(), 9);
Special case, creates a uniform GL_UNSIGNED_BYTE from a pointer.

OpenGL::Array objects are Perl references; in order to use them in OpenGL APIs that expect C pointers, you need to use the native pointer:
my $array = OpenGL::Array->new(4, GL_INT);
glGetIntegerv_c(GL_VIEWPORT, $array->ptr);
my @viewport = $array->retrieve(0, 4);

assign$array->assign($pos, @data);
Sets array data starting at element position $pos using @data.
assign_data$array->assign_data($pos, $data);
Sets array data element position $pos using packed string $data.
retrievemy @data = $array->retrieve($pos, $len);
Returns an array of $len elements from an array object.
retrieve_datamy $data = $array->retrieve_data($pos, $len);
Returns a packed string of length $len bytes from an array object.
elementsmy $count = $array->elements();
Returns the element count from an array object.
ptrptr = $array->ptr(); # typically passed to opengl _c functions
Returns a C pointer to an array object.
Returns a C pointer to an array object.
offsetptr = $array->offset($pos);
Returns a C pointer to the $pos element of an array object.
update_ptr$array->update_pointer($ptr);
Points the existing OpenGL::Array to a different data pointer.

Helps abstract Vertex Array and VBO rendering.
# Requires GL_ARB_vertex_buffer_object extension and POGL 0.55_01 or newer
bind$array->bind($id);
Binds a GPU buffer to an array object. If bound, glXxxPointer_p APIs will call glBindBufferARB.
boundmy $id = $array->bound();
Return bound buffer ID, or 0 if not bound.

Eventually, this API will abstract CPU vs GPU-based affine transforms for the best performance.
affine $array->affine($xform);
# $xform is an NxN OpenGL::Array object used to transform $array.
#N must be one element wider than the width of the array.

calcUsed to populate or mathematically modify an POGL array. Uses Reverse Polish Notation (RPN) for mathematical operations. At the moment, any array used with calc must be made of only of GL_FLOAT types.
$array->calc($value);
Populates the array with $value.
$array->calc(@values);
Populates each row of the array with @values, assuming rows have the same width as the length of @values. If the number of passed values must be evenly divisible by the number of elements in the array. The number of values becomes the number of "columns." The number of "rows" is the total number of elements of the array devided by the columns.
$array->calc(1.0, '3,*', '2,*,rand,+', '');
Resets the first column of each row to 1.0; multiplies the values in the second column by 3; multiplies the third column by 2, then adds a random number between 0 and 1; leaves the fourth column alone. During this particular calc operation there would be 4 columns.
calc maintains a push/pop stack and a "register" for each column.
calc also allows for other OpenGL::Arrays to be passed in. If multiple arrays are passed they must all have the same number of elements. Only the calling array will be operated on, but as each element is visited, the values from the other arrays are pre-added to the stack (in reverse order).
$array->calc($array2, $array3, $array4, @values);
calc currently suports the following primitives:
!Logical "Not" for End of Stack (S0) for the current column; becomes 1.0 if empty or 0. otherwise 1.0
-Arithmetic Negation of S0
+Add S0 and Next on Stack (S1), pop operands and push result (Result)
*Multiply S0 and S1; Result
/Divide S1 by S0; Result
%S1 Modulus S0; Result
=Test S0 equality to S1; pop operands and push non-zero (1.0) for true, otherwise 0.0 (Boolean)
>Test if S0 Greater than S1; Boolean
<Test if S0 Lesser than S1; Boolean
?If S0 is true (non-zero), pop S0 and S1; otherwise pop s0-3, push s1
popPop s0
randPush a random number from 0.0 to 1.0
dupPush a copy of S0
swapSwap values of S0 and S1
setCopy S0 to the column's Register
getPush the column's Register onto the column's Stack
storePop S0, and copy the values from the matching row of the passed OpenGL::Array at that index. Values are copied into the current column registers.
my $o1 = OpenGL::Array->new_list(GL_FLOAT, 1, 2, 3, 4, 5, 6); my $o2 = OpenGL::Array->new_list(GL_FLOAT, 7, 8 ,9, 10, 11, 12); $o1->calc($o2, "1,store,get","","get"); $o1->retreive(0,6) will be (7, 2, 9, 10, 5, 12)
loadPop S0, and set the values of the matching row of the passed OpenGL::Array named at that index. Values are copied from the current column registers.
my $o1 = OpenGL::Array->new_list(GL_FLOAT, 1, 2, 3, 4, 5, 6); my $o2 = OpenGL::Array->new_list(GL_FLOAT, 7, 8 ,9, 10, 11, 12); $o1->calc($o2, "set","", "set,1,load"); $o2->retreive(0,6) will be (1, 0, 3, 5, 0, 6)
colgetPop S0, and push the column S0 value onto the current stack.
$o = OpenGL::Array->new_list(GL_FLOAT, 1, 2, 3, 4, 5, 6);
$o->calc('2,colget','','');
# $o->retreive(0,6) will be (3, 2, 3, 6, 5, 6)
colsetPop S0, and set the column S0 value to the new top of the stack.
$o = OpenGL::Array->new_list(GL_FLOAT, 1, 2, 3, 4, 5, 6);
$o->calc('27,2,colset','','');
# $o->retreive(0,6) will be (1, 2, 27, 4, 5, 27)
rowgetPop S0 and S1, and push the column S0 value from row S1 onto the current stack.
$o = OpenGL::Array->new_list(GL_FLOAT, 1, 2, 3, 4, 5, 6);
$o->calc('1,2,rowget','','');
# $o->retreive(0,6) equiv (6, 2, 3, 6, 5, 6)
rowsetPop S0 and S1, and set the column S0 value of row S1 to the new top of the stack.
$o = OpenGL::Array->new_list(GL_FLOAT, 1, 2, 3, 4, 5, 6);
$o->calc('27,1,2,rowset','','');
# $o->retreive(0,6) will be (1, 2, 3, 4, 5, 27)
endEnd processing; column unchanged
endifPop S0, End if true; column unchanged
endrowEnd processing of current row; column unchanged
endrowifPop S0, End processing of current row if true; column unchanged
returnEnd processing; column value set to s0
returnifPop S0, End if true; column value set to s0
returnrowEnd processing of current row; column value set to s0
returnrowifPop S0, End processing of current row if true; column value set to s0
ifalias to ?
oralias to +
andalias to *
incAdd 1 to S0
decSubtract 1 from S0
sumAdd and pop everything in stack; push result
avgAverage and pop everything in stack; push result
absReplace S0 with its absolute value
powerRaise S1 to the power of S0; Result
minThe lower of S0 and S1; Result
maxThe higher of S0 and S1; Result
sinSine of S0 in Radians; Result
cosCosine of S0; Result
tanTangent of S0; Result
atan2ArcTangent of S1 over s0; Result
countPush the number of elements in the array
indexPush the current element index (zero-based)
columnsPush the number of columns in the array
columnPush the current column index
rowsPush the number of rows in the array
rowPush the current row index
piPush the the value of PI (but remember calc is just for floats)
dumpPrint a dump of the current stack to standard out.
OpenGL::Array->new_list(GL_FLOAT,7)->calc("dup,dec,2,swap,10,4,set,dump");
Would print:
-----------------(row: 0, col: 0)----
Register: 4.0000000
Stack 4: 7.0000000
Stack 3: 2.0000000
Stack 2: 6.0000000
Stack 1: 10.0000000
Stack 0: 4.0000000

Bulk of documentation taken from http://graphcomp.com/pogl.cgi?v=0111s3p1&r=s3p6
Additions by Paul Seamons