1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
|
In this section we specify tt(using IntVect = std::vector<int>) to illustrate
the construction of an expression template.
Starting point is a simple tt(main) function, in which several tt(IntVect)
objects are added. E.g.,
verb( int main()
{
IntVect one;
IntVect two;
IntVect three;
IntVect four;
// ... assume the IntVects somehow receive values
four = one + two + three + four;
})
At this point the code does not suggest that expression templates are
going to be used. However, tt(operator+'s) implementation is special: it's a
template merely returning an object constructed by tt(operator+):
verb( template<typename LHS, typename RHS>
BinExpr<LHS, RHS, plus> operator+(LHS const &lhs, RHS const &rhs)
{
return BinExpr<LHS, RHS, plus>{ lhs, rhs };
})
Our expression template is called tt(BinExpr). It has three template type
parameters: two object types and a template template parameter performing the
requested operation. Its declaration looks like this:
verb( template<typename LHS, typename RHS, template<typename> class Operation>
struct BinExpr;)
Since tt(LHS) and tt(RHS) can either be the data type that is processed by
the expression template, or a tt(BinExpr) two different typenames are
required. tt(Operation) is the operation that is performed by the expression
template. By using a template template parameter we can use tt(BinExpr) to
perform any operation we want, not just addition. Predefined function
templates like tt(std::plus) can be used for the standard arithmetic
operators; for other operators we can define our own function templates.
tt(BinExpr's) constructor initializes constant references to tt(lhs) and
tt(rhs). Its in-class implementation is
verb( BinExpr(LHS const &lhs, RHS const &rhs)
:
d_lhs(lhs),
d_rhs(rhs)
{})
To retrieve the resulting tt(IntVect) a conversion operator is defined. We
already encountered its implementation (in the previous section). Here is it,
as an in-class implemented tt(BinExpr) member:
verb( operator ObjType() const
{
ObjType retVal;
retVal.reserve(size());
for (size_t idx = 0, end = size(); idx != end; ++idx)
new(&retVal[idx]) value_type((*this)[idx]);
return retVal;
})
We return to the type tt(ObjType) below. At this point it can be
considered an tt(IntVect). The member tt(size()) simply returns
tt(d_lhs.size()): in any sequence of tt(IntVect) additions tt(LHS) eventually
is an tt(IntVect), and so every tt(BinExpr) defines a valid tt(size()) like
so:
verb( size_t size() const
{
return d_lhs.size();
})
The only remaining member to implement is tt(operator[]). Since it
receives an index, it only needs to perform the requested operation on the
corresponding index elements of its tt(d_lhs) and tt(d_rhs) data members. The
beauty of expression templates is that if either one itself is a tt(BinExpr)
that expression template in turn calls its tt(operator[]), eventually
performing the requested operation on all corresponding elements of all
tt(IntVect) objects. Here is its implementation:
verb( value_type operator[](size_t ix) const
{
static Operation<value_type> operation;
return operation(d_lhs[ix], d_rhs[ix]);
})
This implementation uses another type: tt(value_type) which is the
type of the elements of the vector type that is processed by the expression
template. Like tt(ObjType) before, its definition is covered below. The static
data member tt(operation) simply is an instantiation of the tt(Operation) type
that is specified when constructing an tt(ExprType) object.
In the next section we take a closer look at tt(ObjType) and tt(value_type).
|
