[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20211022170732.d7lg6i55m6wo2x5p@chazelas.org>
Date: Fri, 22 Oct 2021 18:07:32 +0100
From: Stephane Chazelas <stephane@...zelas.org>
To: Oğuz <oguzismailuysal@...il.com>
Cc: Rob Landley <rob@...dley.net>, enh <enh@...gle.com>,
Rich Felker <dalias@...c.org>,
austin-group-l <austin-group-l@...ngroup.org>,
"libc-coord@...ts.openwall.com" <libc-coord@...ts.openwall.com>
Subject: Re: Posix issue 8 pending change to system().
2021-10-22 12:43:11 +0300, Oğuz:
[...]
> > So system() was broken when sh started accepting more than one
> > option argument.
> >
>
> I wouldn't say broken. This is rather an academic case, I don't see why
> anyone would name a utility that way (`-potato'/`+potato'), I don't know a
> single utility that is named that way either.
[...]
FWIW,
zsh has a "-" builtin command (to run a command with it's
argv[0] prefixed with a "-").
$ zsh -c '- ps -f'
zsh: bad option string: '- ps -f'
$ zsh -c -- '- ps -f'
UID PID PPID C STIME TTY TIME CMD
chazelas 728758 93970 0 18:01 pts/8 00:00:00 /bin/zsh
chazelas 729347 728758 0 18:02 pts/8 00:00:00 -ps -f
(from 1990; less useful now that most shells support a -l /
--login to achieve the same effect as "- sh").
Note that it's not only about the command's name, one could want
to do:
system("++dir++/script.py blah");
> The standard should note this corner case and encourage developers to
> implement `system()' to behave as if `sh -c -- command' were called, and
> leave it at that.
[...]
Yes, it should make sure that system("any shell code") correctly
gets a shell to interpret that code.
--
Stephane
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
