To: corbet-lwn@eklektix.com
Subject: Linux Standard Base session reports
From: Jonathan Corbet <corbet@eklektix.com>
Date: Wed, 03 Mar 1999 21:46:57 -0700

Dan Quinlan gave a talk on the status of the Linux Standard Base project.
Here's a minimal summary.  His slides will evidently appear on the LSB web
pages (http://www.linuxbase.org/).

Much of the talk was a rehash of what the LSB is, and why.  See their web
site for this information.  Dan did point out the libc/glibc difficulties
as one example of why the LSB is needed.

The LSB will be done in three pieces:

- A written specification
- A test suite
- A sample implementation

Progress is happening on the first two, the sample implementation isn't
going anywhere yet.

Initially they are working on the Intel architecture only; other hardware
will be brought in later.

The specification includes X11 libraries.  It was asked whether systems
without X would be compliant.  Answer is "no."

The project must break new ground in some areas where there is no existing
standard practice.  Example was packages that need to add a user account.
Adding accounts is done differently on different distributions, and none
really address the case of large networks where accounts are not controlled
locally.  That one will be an interesting challenge.

The test suite will be released as "almost free" software.  It's not cool
to modify the tests to make your system pass.  There will also be a tester
program for applications to insure that they are compliant.

The status: the written specification is in progress, probably requires
another six months for some sort of completion.  Some parts of the test
suite are in beta.  They will not be making "releases" of the LSB, since
they don't want anybody to try to try to claim compliance with anything but
a final version.

Active participation is being seen from Caldera, Red Hat, Debian, and
SuSE.  There is interest from Oracle and Informix, though they are still
mostly in a "wait and see" mode.

They are trying to avoid the whole GNOME/KDE thing.  Their rule of thumb is
that if 95% of people are not already using something, it's probably not a
good thing to try to standardize.