[LWN Logo]

Date: Tue, 19 Jan 1999 23:53:17 -0800
From: Daniel Quinlan <quinlan@transmeta.com>
To: Joseph Carter <knghtbrd@debian.org>
Subject: Re: LSB?

Joseph Carter <knghtbrd@debian.org> writes:

> Reasonable objection notwithstanding, I intend to write a letter to those
> responsible for the LSB to attempt to raise the issues we have with their
> current proposal.  I would appreciate discussion on these issues in other
> parts of this thread.  I encourage those who have a significant opinion
> not yet voiced in the LSB thread found on debian-devel to write them down
> either as part of the thread or directly to me to aid in the drafting of
> this letter.

I'm jumping in the discussion a little late (I just joined this list),
but please let me try to help explain things...

I'd like to fix the problems that Debian developers are finding in the
LSB.  I think most of the i386isms are due to problems in FHS (you can
blame me), which will be fixed in FHS 2.1.  Remember that the original
FHS dates back to when i386 was the only architecture included in
Linus' kernel. (Patches to the FHS source are welcome.)

I also imagine that some people have some concerns with the TOG FHS
test suite.  Basically, anyone is free to make a contribution to the
LSB test suite effort -- provided that:

  - It's free ("Open Source") and released under the license we say it
    should use.  (Since we haven't chosen that license yet, Andrew Josey
    is using the Artistic license for now, but he agreed to switch when
    the LSB makes that decision.)
  - It must be in sync with the LSB written spec (which references the
    FHS) and the LSB sample implemention should pass it.
  - It won't be incorporated into LSB 1.0 without passing muster of
    the LSB test suite group (headed by Dale Scheetz).

The technical problems you note are due to deficiencies in the written
specification (in FHS 2.0), and are not mistakes on the part of Andrew
Josey of the Open Group.

Andrew has contributed more to the LSB effort than most people.  BUT,
the TOG is *not* defining LSB.  Linux people are defining it -- and if
a company passes every hurdle we insist that they pass, why shouldn't
we allow them to help?  (I haven't seen any marketing spin from TOG,
but if there is any, please point me to it.)

If anyone has interest in helping develop LSB test suites (or other
parts of the LSB), please email me and Dale Scheetz <dwarf@polaris.net>.

- Dan

(By the way, before taking on other projects, I was heavily involved
in Debian, from the early days with Ian Murdock through Bruce Perens.)


--  
To UNSUBSCRIBE, email to debian-devel-announce-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org