[Pkg-utopia-maintainers] Bug#562811: further fudging stages may only confuse things

Mark Hedges hedges at scriptdolphin.com
Tue Feb 9 03:50:29 UTC 2010



On Tue, 9 Feb 2010, Michael Biebl wrote:
>
> > However, I would guess it is still more useful for most
> > packages to run if-down.d scripts instead of
> > if-post-down.d, if only one of them is going to run.
>
> Not exactly sure what you mean by that. Would you please elaborate.

I see.  Maybe I'm the one confounding it.  DWIS (ifupdown
does what NM says) vs. DWIM.

> > To me it seems like "up" and "down" have some sort of
> > primacy over "pre-up" and "post-down."
>
> Might be. Imho I find it a bit confusing, that ifupdown does not map its
> events/PHASES to directories of the same name, i.e.
> PHASE:pre-up -> if-pre-up.d
> PHASE:post-up -> if-up.d
> PHASE:pre-down -> if-down.d
> PHASE:post-down -> if-post-down.d

I think that's the way ifupdown works, see `man 5
interfaces` it is the phase environment variable.
"post-up" is alias for primary entity "up" and "pre-down"
for "down," AFAICT.  I would think NM should set PHASE
values "pre-up", "up", "down", and "post-down" since those
are the established directory names.

> At the current state it is imho rather unlikely that
> upstream is changing his position on this.

Yargh.  Argh.  Avast ye.

It's anti-competitive for one engineer of a particular
commercial distribution to make this decision.  (See below.)

> I'm a bit puzzled why you think that keeping down mapped
> to pre-down is preferrable.

Because it only requires changing the way one package works
(network-manager) versus two (network-manager and ifupdown).
The scripts installed with those directory names (which are
useable as PHASE values) expect to run in a universal order
regardless of whether the interface is managed by
network-manager or spelled out in interfaces.

Debian is a model for software distribution free of economic
encumberances, so it represents and equates to free access
to the entire market.  Companies that produce and sell
specialty software packaged for use in a "pure" open source
environment are the only ones who truly have access to the
entire market.  I could imagine some product to sell, like a
cell phone gui, that would bring interfaces up and down; a
manager for active connection routes switching from one
wireless to another might expect to deploy scripts to send
route data precisely when new interfaces came up and before
old ones went out of range.

Say I developed that with Debian, because that platform
technically and legally grants me access to the whole world
market, since the platform has no cost to the consumer.
Software can be ported to RedHat, Mac and Windows easily
enough; it's a growing model.  It's not fair to consumers
using the free platform or the other operating systems for
one company to take over community development of a
secondary system (NM) which utilizes a core system
(ifupdown's $PHASE, also can be utilized by other secondary
network managers), and then leverage that development to
disenfranchise them from their products of choice.

> -- Why is it that all of the instruments seeking
> intelligent life in the universe are pointed away from
> Earth?

Because they haven't detected higher-dimensional gravity
reverberation transmissions, so they don't know to look?

Mark





More information about the Pkg-utopia-maintainers mailing list