RT++ --- Tutorial and Reference Manual -- Definition

$\"\"$ $\"\"$ $\"\"$
Go backward to Representation
Go up to Threads
Go forward to Basics

Definition

Thread<R> t
Thread $n$ <R, A₁, ..., A_n> t
Thread $n$ <R, A₁, ..., A_n> t(f, a

₁, ..., a

_n)
R (*f)(A₁, ..., A_n)
A_i a

Specification: Defines a thread handle t.

In the first form, t may be bound to any thread that returns a result of type R. t is not yet bound to any thread.
In the second form, t may be bound to any thread that takes $n$ arguments of types A_i and returns a result of type R. t is not yet bound to any thread.
In the third form, t is bound to a new thread that returns f(a₁, ..., a_n).

Note: A thread handle of type Thread $n$ <R, A₁, ..., A_n> is implicitly converted to a thread handle of type Thread<R> whenever such a type is expected. Thus e.g. the following is legal:

Thread<R> s
Thread $n$ <R, A₁, ..., A_n> t(f, a₁, ..., a_n)
s = t

Restrictions: The number of arguments

n

ranges from

0

to an implementation-dependent constant (currently

4

). This range may be extended by defining the corresponding template class.

Implementation: The third form is more efficient than the otherwise equivalent form

Thread $n$ <R, A₁, ..., A_n> t
t = Thread $n$ <R, A₁, ..., A_n>(f₁, a₁, ..., a_n)

Normally this form is also equivalent to to

Thread $n$ <R, A₁, ..., A_n> t
t.create(f₁, a₁, ..., a_n)

However, if the preprocessor constant RT_THREAD_NOTLAZY is defined before including the header file rt++.h, the declaration is equivalent to

Thread $n$ <R, A₁, ..., A_n> t
t.start(f₁, a₁, ..., a_n)

Author: Wolfgang Schreiner
Last Modification: April 12, 1997