Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!POSTGRES.BERKELEY.EDU!dillon
From: dillon@POSTGRES.BERKELEY.EDU (Matt Dillon)
Newsgroups: comp.sys.amiga.tech
Subject: Re: fork() (not forkv()!) on Amiga?
Message-ID: <8910110036.AA13975@postgres.Berkeley.EDU>
Date: 11 Oct 89 00:36:06 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Lines: 59

:Here's another moderately related question on executing programs.
:I need to call a program from within mine.  I've found several ways
:to do this:
:
:    1) LoadSeg, CreateProc, UnloadSeg
:    2) Lattice forkv() or forkl()
:    3) Lattice system()
:    4) Execute()

	One thing to note:  In OS1.4, all C: programs will be written
    in C.  Thus, Lattice's fork*() calls should work just fine then.

	I talked to John Toebes 'bout this.  They didn't make fork*()
    compatible w/ BCPL programs because you need to do a lot of
    unsupported structure munging to setup the global vectors right.
    So Lattice is using a more supportable method with its fork*()
    calls at the expense of BCPL compatibility.

:Unfortunately they each seem to have problems.  The Lattice forkv()
:call causes a GURU when it tries to execute almost anything in the c:
:directory.  It also GURU's on a few other things as well - this
:behaviour makes its use unacceptable.
:
:Lattice system() call will not accept input and output file streams
:to let me redirect the output or input of a command.
:
:Execute() comes close, but I can find no way to obtain the return code
:of the executed command.  The Execute() call itself always returns
:-1, and IoErr() always contains zero.  I need to be able to examine the
:return code.

	Was the Execute() return code problem going to be fixed in
    1.4?  That and properly disconnecting the console device.  Rightly,
    if RUN detects <NIL: >NIL: it ought to fill in pr_ConsoleDevice
    with a dummy console (like NULL:)

:LoadSeg, CreateProc, UnloadSeg has no apparent way to pass command line
:parameters to a command.  Also, I've heard that it will also crash on
:c: programs.

	CreateProc() has no apparent way to pass command line params
    to the command.  You need to CreateProc() to a tag that sets up 
    all the process fields properly and then there is the question of
    what to do about the CLI pointer...

:So, how do you do it?  There must be an easy way to execute a program
:that won't crash on some programs, and will let you have the return
:code back from the command, and lets you redirect I/O.
:
:              - thanks
:
:                  steve (koren@hpfela.HP.COM)

	1.4 is really comming together... it isn't vapor and developers
    have received 1.4 alpha 16 for testing... it's pretty solid.

	I think the issue will not be resolved until 1.4 comes out.

				-Matt