Re: Bugs in bash (was: Release-critical Bugreport)
- To: firstname.lastname@example.org
- Subject: Re: Bugs in bash (was: Release-critical Bugreport)
- From: Steve Greenland <email@example.com>
- Date: Sat, 29 May 1999 13:52:12 -0500
- Message-id: <19990529135212.A19818@molehole>
- Mail-followup-to: firstname.lastname@example.org
- Reply-to: Steve Greenland <email@example.com>
- In-reply-to: <19990529182254.A1584@portaloo.>; from Edward Betts on Sat, May 29, 1999 at 06:22:54PM +0100
- References: <19990528001503.A9598@master.debian.org> <19990529084332.B1606@portaloo.> <19990529120021.A19542@molehole> <19990529182254.A1584@portaloo.>
On 29-May-99, 12:22 (CDT), Edward Betts <firstname.lastname@example.org> wrote:
> The other question does Debian support or advocate the changing of the /bin/sh
> link? There are still lots of scripts that need bash but say they need sh
> could break some things.
This comes up every 6 months or so and inevitably devolves into a huge
flame war. The general consensus seems to be:
1. /bin/sh must be a POSIX compliant shell.
2. Any script that says "#!/bin/sh" but uses bash-specific constructs
has a bug in it.
3. /bin/bash must be present.
(The third item seems to be the one that annoys the most people.)
As far as I know, no one has replaced /bin/sh with something besides
bash (ash, for example) *and* reported back sufficiently positive
results to allow us to change the default /bin/sh to something smaller
than bash. Given that bash must be present on the system, the only clear
incentive to change /bin/sh to something smaller is start up time, which
is non-trivial during the boot process.o
I think that's a fair summary of the bash vs. sh situation.
Theoretically, we support non-bash /bin/sh, but I'd hardly say we
advocate it. As a practical matter, I'm sure there are a bunch of
standard scripts that are screwed up. I'm even more positive that on
a true multi-user system, there are tons of user written scripts that
assume /bin/sh is bash.