Re: POSIX shell; bash ash pdksh & /bin/sh
>>"Santiago" == Santiago Vila <email@example.com> writes:
Santiago> On 3 Aug 1998, Manoj Srivastava wrote:
>> The essentialness of bash is not quite
>> irreversible, but it is hard enough that it requires good reasons to
>> remove. You have provided none that convinced me.
Santiago> Manoj, the problem with this discussion, I think, is that
Santiago> you use the word "essential" with two different meanings,
Santiago> 1. That bash has the essential flag in the control file.
Santiago> 2. That bash is a very important package and we should
Santiago> never never never let the user to remove it from their
Santiago> system under any circumstance even if there is a posix
Santiago> /bin/sh replacement, not today nor tomorrow nor the year
Santiago> 2000 nor the year 2010.
Actually, without thought, I would say never. You have to have
convincing reasons to remove bash, which I have not yet seen.
Santiago> and mix the two meanings constantly.
Santiago> I am *always* talking about essentialness in the first meaning.
And I say we need to think long and hard about removing the
essential flag from a package, becuse that step is going to be costly
(figuring out the implicit dependencies and adding stuff to the
depends line). I am questioning whether it is cost effective to do
that. (That does not mean virulent opposition; I just need to be
convinced it is worth while jumping through these hoops).
Santiago> The fact that 90% of scripts *already* use /bin/sh and not
Santiago> /bin/bash *proves* that bash is not as essential (in the
Santiago> *second* meaning) as you say, and therefore making it
Santiago> non-essential (in the first meaning) is, at least,
Santiago> something we should consider the day we have another
Santiago> /bin/sh other than bash.
I am for *testing* other packages to serve as /bin/sh; but I
also believe that in any package I should be able to put scripts that
say #!/bin/bash and expect things to work; including preinsts
(essential packages are all I can depend on, in preinst, I am told).
I think I would need to see alternates tested even before I
move from this position; and I would need to have a POSIX compliant
shell that I can *depend* to have available, which has foibles that
can be trusted; and not just use /bin/sh that can be any shell at all
(within guidelines that we set) and use just the lowest common
denominator (which we all agree seems to be POSIX). Even POSIX
compliant shells behave differently on different scripts; and I think
it is nice we have one command nterpreter with well known behaviour.
Date: 21 Feb 90 20:51:25 GMT From: firstname.lastname@example.org (Randal
Schwartz) echo Just another Perl hacker,|perl -e
Manoj Srivastava <email@example.com> <http://www.datasync.com/%7Esrivasta/>
Key C7261095 fingerprint = CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
To UNSUBSCRIBE, email to firstname.lastname@example.org
with a subject of "unsubscribe". Trouble? Contact email@example.com