home *** CD-ROM | disk | FTP | other *** search
- Path: oberon.nbg.sub.org!not-for-mail
- Newsgroups: comp.sys.amiga.programmer
- References: <4ju349$r1e@sparky.navsea.navy.mil> <4jvrqs$hk0@btmpjg.god.bel.alcatel.be> <slN3BCS00iWZ0_8G1R@andrew.cmu.edu> <316533DF.5BFD@bat710.univ-lyon1.fr>
- From: "hartmut Goebel" <hartmut@oberon.nbg.sub.org>
- Date: Fri, 05 Apr 1996 18:37:42 +0100
- X-NewsReader: IntuiNews 1.3b Beta 12 (23.3.96)
- Subject: Re: Messages vs. Semaphores for external clocking
- Message-ID: <1779969@oberon.nbg.sub.org>
-
- Laurent GIROUD (milu28@bat710.univ-lyon1.fr) schrieb:
-
- > Of course but you must not forget something -> in order to be sure
- >that
- >the semaphore you are just refering to in your program is valid you
- >actually HAVE to FindName() it before and then to acces it in an atomic
- >operation thus needing to embrace the two instructions (FindName +
- >ObtainSemaphore) in a Forbid()/Permit() pair...
-
- No! This is only necessary if the Semaphore can go away anytime. If you
- implement some protkoll using something like client-registration, this can
- not happen. Thus the Semaphore will stay vaild until the clients says goodby,
- thus no FindName() needed everytime, thus no Forbit/Permit.
-
- > This could be added to "The OS improvement list" :) ...
-
- Unnecessary overhead!
-
- >Laurent Giroud
-
- +++hartmut
-
- | hartmut Goebel | hartmut@oberon.nbg.sub.org // |
- | Software Manufactur | Essich@irc \X/ |
-