Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site amiga.amiga.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!decvax!decwrl!pyramid!amiga!robp
From: robp@amiga.UUCP (Robert A. Peck)
Newsgroups: net.micro.amiga
Subject: Re: Processes and return values
Message-ID: <979@amiga.amiga.UUCP>
Date: Thu, 10-Apr-86 19:57:44 EST
Article-I.D.: amiga.979
Posted: Thu Apr 10 19:57:44 1986
Date-Received: Sat, 12-Apr-86 22:45:32 EST
References: <11701@watnot.UUCP> <974@amiga.amiga.UUCP>
Reply-To: robp@dudley.UUCP (Robert A. Peck)
Distribution: net
Organization: Commodore-Amiga Inc., 983 University Ave #D, Los Gatos CA 95030
Lines: 67

Responding further to Matt and Rico .... AmigaDOS has been flamed for
the way it does certain things and people have, along the way, either
found or been given the tools that are needed to perform various
operations in a more suitable ( Un*x-like?) manner.  People who write
things like CLI's most probably will be building in various command
ability rather than asking the DOS to load the command such as type,
copy, and so on.  I venture to say that there will be very (very very very???)
few applications released based purely on BCPL and developed with the
full background of knowing AmigaDOS-related TRIPOS process internals,
since the experts therein are very few and far between.  So why not 
concentrate on applications for and in the languages with which we are
all most familiar.

**** small flame-on *****

Why should it be necessary to spend the X-pages of code to set up the
BCPL process model and associated CLI data structures when, by a small
amount of code as I've already released, one can run most any program
without any of the CLI overhead.   It is simply unnecessary for the 
small amount of return one gets.  

**** small flame-off ****

A second thing to note... only if a program ends by "exit(value)" rather
than "return(value)" can that value be sensed ..... EVEN BY THE CLI!
(See the Amiga C startup code:  Astartup.asm .... little piece of it 
here to show why this happens... so a program has to be intelligent 
enough in the first place to be capable of "returning" a code).   So 
any program that uses the standard C startup has a choice of whether 
it can or cannot return a code to the starting process or CLI.

 
	domain:  
       	         jsr     _main
					 ; arrives here if return(value)
					 ; from _main.  Notice that the
					 ; value is ignored.
       	         moveq.l #0,d0           ; Successful return code
				 	 ; See, return code = 0 no matter
					 ; what value is return()'ed.	
       	         bra.s   exit2
	_exit:
       	         move.l  4(SP),d0        ; extract return code from stack
					 ; because the C code translated
					 ; exit(value) as a function, that
					 ; pushes value on the stack before
					 ; calling the underlying function
					 ; named _exit.
	exit2:
       	         move.l  initialSP,SP    ; restore stack pointer
       	         move.l  d0,-(SP)        ; save return code
 
		; ****** ( more stuff here ) **************

       	 	;------ this rts sends us back to DOS:
	exitToDOS:
                move.l  (SP)+,d0
                rts

Nuff said.
robp.


Disclaimer:  The opinions expressed above are mine and do not necessarily
	     reflect the opinions of my employer.   But so far C-A hasnt
	     released the source code of the CLI, so I suppose I shouldn't
	     do it either.