-----Original
Message-----
From: Anthony Towns [mailto:aj@azure.humbug.org.au]
Sent:
Wednesday, February 20, 2002 9:18 AM
To: debian-devel@lists.debian.org;
lsb-spec@lists.linuxbase.org
Subject: Re: Bug#134658: ITP: lsb -- Linux
Standard Base 1.1 core
support package
On Wed, Feb 20, 2002 at
09:11:14AM -0600, VomLehn, David wrote:
> I would urge some caution here.
If I were to guess at a reason as to why
> bin has uid/gid of 1, I would
suspect that some application
> installations use 1 as the bin uid/gid
instead of looking it up. This
> may not be good practices, but the LSB
work is primarily focused on
> codifying current practices, good or
not.
No, it's not. The LSB is focussed on codifying practices that can be
used
to install third party software on any distribution of Linux.
Assuming
the uid of the bin user is 1, is not one of those practices, and
it
should never have been codified.
Yes, and I think it's likely we made a mistake. It's now too late to change that for versions 1.0 and 1.1 of the LSB specification and I suggest that we simply understand the current state of affairs, accept it, and implement a reasonable process to fix the mistake. Doing so will take at least a year after the fix process begins to allow any application providers to make the required change.The fix for this follows a well-beaten path--we are not the first standardization body to make a mistake. Assuming there is nobody who can find any reason for the current state of affairs, the very next release of the LSB includes, 1) a statement that the requirement that uid and gid be one is deprecated and will become obsolete in a future release, 2) specify that the plan is to obsolete this requirement in the first approved version of the LSB specification after a particular date, and 3) deliver on item 2.
This change will not make any distribution
non-conformant, and since
there are absolutely no LSB compliant apps at
present, won't break any
of them, either.
Yes, and it's great when we can move forward without breaking anything. The fact that there are no LSB compliant applications at present does not mean that there are no applications which completely meet the LSB specification. When a proper validation procedure exists for formally declaring an application compliant with LSB 1.1, the ones which assume that the uid and gid of bin are one will also be compliant. Since I expect the LSB effort to continue into the future, I also expect that some LSB 1.1 compliant applications will not be LSB compliant to a future version of the specification.
I think I've made my position clear, so I have nothing further to discuss on this issue in LSB mailing lists. If anyone else wishes to engage me in further dialog, please send email directly to me. If there is a reason to view things differently, I'll publicize it myself.