(cc:ed to #218530)
Guilherme de S. Pastore wrote to #289732:
> The attached patches make your maintainer scripts behave just like a
> package using debhelper would
The patch makes use of "which". The problem with "which" is supposed to
be that it is in /usr/bin/ and (as Ian Jackson wrote in #218530):
> `which' is wrong, because we want to know how the shell which is
> actually executing the maintainer script will execute the command.
Andreas Metzler wrote later in #218530:
> FWIW posix suggests to use "command -v" if available, "The command -v
> and -V options were added to satisfy requirements from users that are
> currently accomplished by three different historical utilities: type
> in the System V shell, whence in the KornShell, and which in the C
> shell.[...]"
So I take it that "command -v" is actually recommended. The problems
(judging from the original submission) are, first, that it is being used
in a #!/bin/sh script, whereas "command -v" isn't implemented by all
POSIX shells (see policy 10.4), and second, that the command isn't being
checked for executability.
What is the best way to run a command if and only if it is on the PATH?
Debian policy section 9.3.3.2 was later changed by Manoj to show how to
test for the presence of a command using "command -v".
> if command -v invoke-rc.d >/dev/null 2>&1; then
> invoke-rc.d
> else
> /etc/init.d/
> fi
However, this does not check that the invoke-rc.d file found on the
patch can actually be executed. So I guess you need to do this:
#!/bin/bash
...
if CMD="$(command -v update-menus 2>/dev/null)" && [ "$CMD" ] && [ -x
"$CMD" ] ; then
update-menus
fi
--
Thomas Hood <[EMAIL PROTECTED]>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
