Path: utzoo!mnetor!uunet!cbmvax!ulowell!page From: page@swan.ulowell.edu (Bob Page) Newsgroups: comp.sys.amiga Subject: Re: Commodore confirms 1.3 rumor Message-ID: <2598@swan.ulowell.edu> Date: 2 Feb 88 20:06:54 GMT References: <6825@agate.BERKELEY.EDU> Reply-To: page@swan.ulowell.edu (Bob Page) Organization: University of Lowell, Computer Science Dept. Lines: 43 spencer@eris.berkeley.edu (Randal m. Spencer [RmS]) wrote: >Wait a minute, I don't want people spread rumors that I'm human! Now that's a nasty rumor! :-) >I don't even believe that redirection is handled by the shell (CLI), >I think that is part of the startup code that you link your program >with (but I should check that before I make any wild claims). The CLI sets up the input and output file handles based on your redirection specifiers. If it was in the startup code, you could easily lift the restriction that the redirection qualifiers be the first arguments after the command name. >So if I can only have one command on a line, it would >be hard for me to construct the command "dir | more". It would be easy if the IO were done in the startup code: dir would see the |, exec 'more' with an input file handle the same as dir's output handle, then dir would wait around until more did its thing before both were unloaded. Of course it's a little more complicated than that, but the point is all kinds of things would be possible if the CLI just passed its args to the command's startup code. It doesn't, though -- the CLI does the IO handler startup before the program is invoked. HOWEVER, I agree with Randy's statement that there won't be real (as in recognized/handled by CLI) pipes in workbench 1.3, since it requires a ROM/Kickstart change to modify the CLI. If CBM delivers pipes in WB1.3, it will be in the form of something that lays on top of the console handler, like Conman. Of course, it could be just a device called PIPE: and no support in the CLI, which translates to: run dir >pipe:foo more