Path: utzoo!utgpu!watmath!clyde!att!cord!nsw
From: nsw@cord.UUCP (Neil Weinstock)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Help spawning a simple task
Summary: Solved!
Message-ID: <645@cord.UUCP>
Date: 27 Dec 88 04:47:31 GMT
Organization: The Flying Squid Patrol
Lines: 38

Expires:
References: <644@cord.UUCP> <7070@batcomputer.tn.cornell.edu>
Sender:
Reply-To: nsw@cord.UUCP (Neil Weinstock)
Followup-To: 
Distribution: usa
Organization: The Flying Squid Patrol
Keywords: 

In article <7070@batcomputer.tn.cornell.edu> riley@tcgould.tn.cornell.edu (Daniel S. Riley) writes:
>In article <644@cord.UUCP> nsw@cord.UUCP (Neil Weinstock) writes:
>>I've been unsuccessfully trying to spawn a task.  As soon as I call AddTask()
>>I get a visit from the Guru.  I have been using the RKM (volume I, page 17)
[ ... ]
>First, remember that the usual Manx startup code is not run before your
>entry point.  None of the registers will be setup the way you expect,
>particularly a4, so you have to compile with absolute addressing, or
>else somehow get a4 setup with the right base address (I'm a little vague
>here because I'm a Lattice user...sorry).  Remember, this is an entire
>separate task you're setting up, with very little inherited from the 
>parent.

Bingo.  In fact, the Manx geta4() call turned out to be the fix (thanks Ali!).

"When all else fails, read the man(x)ual."

>Second, Tasks cannot call DOS routines.  Only processes can.  Delay()
                                                               ^^^^^^^
>is a DOS routine, so your call to Delay() in EntryPoint() is definite
 ^^^^^^^^^^^^^^^^

Wow, I had no idea, and confirmed by the RKM no less.  What's Delay() doing
in dos.library?

+ .- -- .. --. .- .-. ..- .-.. . ... .- -- .. --. .- .-. ..- .-.. . ... +
| Neil Weinstock | att!cord!nsw     | "One man's garbage is another     |
| AT&T Bell Labs | nsw@cord.att.com | man's prune danish." - Harv Laser |
+ .- -- .. --. .- .-. ..- .-.. . ... .- -- .. --. .- .-. ..- .-.. . ... +