Re: bashisems in maintainer scripts.
Adrian Bunk <email@example.com> wrote:
> On Wed, Mar 03, 2004 at 12:41:26AM +0100, David Weinehall wrote:
>> Imho it would be a worthy goal in the long run to make bash
>> non-essential (a first step would be to make all maintainer-scripts
>> using /bin/sh posix-compliant, a second-step to make scripts using
>> /bin/bash posix-compliant), but I suppose I'm alone in this.
> Some questions:
> Why do you want to disallow maintainers to use some more convenient bash
> How do you plan to make an essential package non-essential without
> breaking anything?
> Would you really be able to remove bash? Similar to initrd-tools'
> dependency on dash, every package maintainer would always be allowed to
> depend on bash when he wants to use bash in his scripts. This is
> perfectly legal, and any maintainer is allowed to tag a wishlist bug
> asking for using a posix-compliant shell instead of bash as "wontfix".
That is not completely correct, if bash was not essential anymore it
would not be possible to use it in maintainer scripts. (Even a
pre-dependency is not strong enough for .config or .postrm)
I agree that making bash non-essential might simply not be worth it,
it is no mightily big package (Installed-Size: 872 on woody) and does
not have big dependencies. The potential for breakage OTOH in not
negligable. And isn't bash an important part of having GNU userland?
NMUs aren't an insult, they're not an attack, and they're
not something to avoid or be ashamed of.
Anthony Towns in 2004-02 on debian-devel