Path: utzoo!censor!geac!torsqnt!hybrid!scifi!bywater!uunet!sugar!peter
From: peter@sugar.hackercorp.com (Peter da Silva)
Newsgroups: comp.sys.amiga.advocacy
Subject: Re: How to improve Workbench 2.0!
Message-ID: <1991Mar2.143410.18414@sugar.hackercorp.com>
Date: 2 Mar 91 14:34:10 GMT
References: <1991Feb23.203358.27835@sat.com> <1991Feb28.132253.8993@sugar.hackercorp.com> <54303@sequent.UUCP>
Organization: Sugar Land Unix -- Houston, TX
Lines: 46

In article <54303@sequent.UUCP> cseaman@sequent.UUCP (Chris "The Bartman" Seaman) writes:
> Wouldn't it be better to enforce a rule that the "!<filename>[!]" MUST
> immediately follow the leading comment character(s)?

What does that buy you? The shell, remember, doesn't know a comment character
from a hole in the ground. This means the shell still has to know all sorts
of special cases... or it'll execute files like this:

+----------
|Remarkable! I have found a proof...
...

You can't disallow ascii text as comment characters, because there have been
languages that used that sort of convention:

+----------
|REM !BASIC! This is a program...
...


You can't even disallow whitespace:

+----------
|( !Forth! requires the comment leadin be a separate word )
...

> Also, it seems
> unnecessary to keep the "#" as part of the requirement.

I was using #! as an example. It could be any unique string:

+----------
|/* @@command=AREXX@stack=10000@window=CON:0/0/640/200/BARTMAN@ */
...

In this case "@@" is the leadin.

> After all, even under the UNIX convention, the "#" is only there because it
> is the accepted comment character for the shell.  It's the "!<filename>"
> that is important.

Well that might be the historical reasoning, but the string "#!" is actually
handled as a 16-bit magic number.
-- 
Peter da Silva.   `-_-'
<peter@sugar.hackercorp.com>.