On Wed, Feb 20, 2002 at 02:31:20PM -0500, Trond Eivind Glomsr?d wrote:
> They don't care about running on Debian - at least, they shouldn't.
You're confused.
ISV's using the LSB obviously care about running on Debian: they care about
running on as many platforms as possible to increase their potential market
share, either so they can make more money, or so they can be more famous.
What they don't want to do is have to special case us. If they follow
the LSB 1.0 or 1.1 specs, and make any use of the bin=uid 1 clause, then
they'll find they *will* have to special case us, and that will annoy
them, since the LSB's raison d'etre is to avoid that nonsense. Correcting
the spec allows us to avoid annoying them. Not correcting it buys no
one anything.
> If
> they have written a program using this part of LSB 1.0, it shouldn't
> matter if it's Debian, Slackware or Caldera they are running on if
> these distribution claim to implement the standard. If they don't
> claim to implement it, the program isn't expected to run anyway.
We're not designing this spec in a vacuum. There are real people out there
with real needs that we're trying to satisfy. If we're not satisfying
their needs now, we need to change it so we are. Debian is one set of
such people, ISVs who want their products to run on both Red Hat and
Debian are another.
Our aim is to let people say "Sure, my program will run on Linux. Doesn't
matter what flavour. Red Hat, SuSE, Debian, Slackware, heck, even BSD
or Solaris via the compatability layers." We're not aiming to have
people say "It'll run on Red Hat and SuSE. It won't run on Debian,
because although that's Linux, it's not *really* Linux". Or at least,
the people who I've talked to aren't.
> Any revision of the standard shouldn't go through overnight,
> but be in a future revision after being carefully reviewed,
> and a rationale for
> the change should be given (as it should have for being in there in
> the first place).
Sure. All of these things should happen. But they shouldn't be used as an
excuse to delay or block a necessary change for months (like has already
happened on this very issue) when the *original spec* had absolutely
none of these things.
> > We could do this tomorrow, and _no one_ would have any cause to
> > complain.
> Anyone using it (developers, book printers, etc) would have cause.
All those masses of developers shipping LSB 1.0 or 1.1 compliant software?
Or the masses of publishers who've already written whole chapters about
why uid 1 should be the bin user, even though we can't think of even a
sentence to justify it?
> Trond Eivind Glomsr?d
> Red Hat, Inc.
Of course, I suppose Red Hat does have a market incentive to make it
as difficult as possible for Debian to comply with the LSB.
"Oh, no, your products won't run on Debian, because those hacker wannabes
can't manage to comply with simple community developed standards. Better
buy Red Hat instead."
So I suppose someone does have a cause to complain afterall. My mistake.
Cheers,
aj, who'd be much less bitter if the last seven months had resulted in
pretty much anything other than "Oh, but Debian should suffer for
our art too, coz Red Hat did" (Hi, Chris), or "Yes, yes, interesting
point. But we must have a careful transition, and working out what
that would be like is on our todo list, so please don't bother
us now".
--
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
We came. We Saw. We Conferenced. http://linux.conf.au/
``Debian: giving you the power to shoot yourself in each
toe individually.'' -- with kudos to Greg Lehey
Attachment:
pgp33FXvXyxWd.pgp
Description: PGP signature