Hello Przemek
Przemyslaw Czerpak-2 wrote:
>
> I will want to introduce two extensions for MT mode.
>
> 1. allow to use hb_mutexCreate() as static variable initialization.
>
> If we unblock using hb_mutexCreate() as static variable initializer
> (now we can use only fully optimized functions: static s :=
> HB_BITOR(1,2))
> then the same code can be reduced to:
> proc p()
> static s_mtx := hb_mutexCreate()
> [...]
> hb_mutexLock()
> [...]
> hb_mutexUnlock()
> [...]
> return
>
> 2. CRITICAL FUNCTION|PROCEDURE
> - user cannot forgot about unlocking the mutex because it will be
> unlocked
> by HVM automatically
> - HVM will unlock them in all cases also when thread generate exception
> like QUIT or BREAK so it's not necessary to use BEGIN SEQUENCE /
> ALWAYS
> statment as protection, f.e. with above modification I can write code
> like:
> proc p()
> static s_mtx := hb_mutexCreate()
> hb_mutexLock()
> [...do sth...]
> hb_mutexUnlock()
> return
> but if [...do sth...] can generate exception, f.e. RT error then
> I should add protection code which will always unlock the mutex:
> proc p()
> static s_mtx := hb_mutexCreate()
> hb_mutexLock()
> begin sequence
> [...do sth...]
> always
> hb_mutexUnlock()
> endsequence
> return
> With CRITICAL functions/procedures to reach the same effect it will
> be enough to make:
> critical proc p()
> [...do sth...]
> return
>
> Both modifications are similar but they are not fully replaceable so
> I want to introduce both.
>
Perfect.
CRITICAL PROC p() is the easiest way. Can it be implemented in C also? Like
:
HB_FUNC_CRITICAL( SOMEFUNC )
{
// do sth
}
Regards
Pritpal Bedi
--
View this message in context:
http://www.nabble.com/MT-extensions-tp20230982p20233603.html
Sent from the Harbour - Dev mailing list archive at Nabble.com.
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
