[boost] Re: Enhanced call_once()

From: Alexander Terekhov (terekhov_at_[hidden])
Date: 2002-08-06 16:44:53

Pete Becker wrote:
[...]
> Of course, multi-threaded code that runs threads from constructors is
> inherently evil, so this really shouldn't be much of an issue in practice.

http://www.terekhov.de/mythread.c

"....
 // i think that the following closely matches Java
 // model with respect to thread objects ("automatic" 
 // lifetime management of thread objects with 
 // optional/multiple/concurrent "join" and creation 
 // of initially suspended threads with "start" method 
 // to resume initially suspended thread and start 
 // execution of user code; call user provided thread
 // start routine)
 // ----
 // i think the later could be useful in C++ library 
 // as well, e.g:
 //
 // // User:
 //
 // mutex mtx;     // Dynamic init
 // condition cnd; // Dynamic init
 //
 // int f1( int,.... ); // Uses mtx/cnd
 // int f2( int,.... ); // Uses mtx/cnd
 //
 // thread_ref< int,.... > tr1( f1, <parms> ); // Dynamic init; BUT library will enforce suspension
 // thread_ref< int,.... > tr2( f2, <parms> ); // Dynamic init; BUT library will enforce suspension
 //
 // // Library:
 // int main( int argc,char* argv[] )
 // {
 // 
 //   // Adopt main thread
 //   //...
 // 
 //   // resume threads created (with 
 //   // suspension enforced) before "main" 
 //   //...
 // 
 //   main_thread::main( argc,argv );
 // 
 //   pthread_exit( NULL );
 // 
 //   return 0;
 // 
 // }
 // 
 // // User:
 // void main_thread::main( int argc,char* argv[] )
 // {
 // 
 //   // ...
 // 
 //   if ( error )
 //     exit( -1 );
 // 
 //   // return from main_thread::main does *not* invoke 
 //   // process termination! main_thread::main behaves 
 //   // just like any other thread function. process 
 //   // termination occurs ether explicitly (exit/...) 
 //   // or implicitly on last thread termination.
 // 
 //   exit( tr1.join() + tr2.join() )
 ...."

regards,
alexander.

P.S. http://www.terekhov.de/thread_ptr.txt ;-) ;-)