:: Re: [DNG] Why Debian 8 Pinning is (…
Top Page
Delete this message
Reply to this message
Author: Rick Moen
Date:  
To: dng
Subject: Re: [DNG] Why Debian 8 Pinning is (or isn't) pointless
Quoting KatolaZ (katolaz@???):

> I personally don't see the reason for such a reaction on your side
> Rick (BTW, welcome here :)), but I am sure I am missing something. I
> personally believe that all the work to avoid and contain systemd and
> other nonsense is valuable, independently of where it comes from :)


I cannot imagine why you would ever think I don't agree!

I can only guess that you didn't bother to attentively read what I said
-- because I think Devuan's work is valuable and have already said so on
my OpenRC conversion page, on the referenced SVLUG mailing list thread,
and on this mailing list.

When you try to pick a fight with someone who admires what you're doing,
just because he doesn't agree with you on every single particular, and
_particularly_ when you attribute to him views he doesn't hold and whose
opposite he just expressed, maybe you should relax a bit and consider
switching to decaf. ;->


> It is true that the pinning game can actually work for you or for me,
> who are content with a minimal window manager and xterm....


I do not call window managers 'minimal'. FWIW, I personally like Window
Maker, in part because it's designed in imitation of NeXTStep's aesthetics
and design, and I was a big fan of NeXTStep back in proprietary Unix
days.

Moreover, althrough I have no personal use for the graphtical
file-manager applications that are the characteristic feature of DEs,
all of them, Thunar, Nautilus, etc., my survey of all Debian
8 'Jessie' packages using apt-caache found _no_ packaged graphical
file-manager app to suffer systemd dependency as presently packaged, not
even GNOME's.

Moreover, if my testing was correct, six DE metapackages (tasks, IIRC)
can be installed in their entirety in Debian 8 'Jessie', and all five of
the others can be installed almost complete, excepting a couple of apps
packaged with a dependency chain to systemd. GNOME, MATE, and Cinnamon
are (naturally) the worst-affected DE metapackages.

What you call 'a minimal window manager and xterm', I see as absolutely
free access to around 20,000 packages including the richest variety of
graphical desktop applications on the planet including (if I cared)
about a dozen graphical file-managers.

Even though I've always thought DEs are bullshit, my results suggest
that a systemd-free Debian 8 'Jessie' system can, with no other
measures, have access to just about everying in all eleven packaged DEs,
and literally the entirety of six of those eleven.

You call that 'minimal'? Seriously? Some minimal!


> ...but this is not a viable long-term solution for the Linux
> community.


First of all, your term 'solution for the Linux community' is
suspiciously vague and undefined.

Second, even if it were defined, I have little confidence that I would
care. It sounds suspiciously like a figurative football to kick around
for polemical purposes. I'm suddenly reminded of any number of
time-wasting threads on comp.os.*.advocacy.

Third, I said nothing on my Web page, the SVLUG mailing list thread,
this mailing list, or anywhere else about providing (or seeking) a
'viable long-term solution for the Linux community' or doing anything
else other than investigating and documenting what happens on Debian 8
'Jessie' when one replaces systemd with a different init system,
enforces a no-systemd package policy using apt thereafter, and observes
what can and cannot be installed from regular packages. FWIW, I found
the few problems resulting to be trivial and orders of magnitude less
numerous than _both_ anti-systemd and pro-systemd people had confidently
predicted.

Additionally, I outlined all the methods I know to work around package
dependency problems that I'm aware of on deb-based architectures -- in
case someone actually cares, e.g., about daisy-player on Debian-stable.

_And_, I carefully qualified what I said about use-cases that this
approach could address only with some difficulties.

Finally, I added my page to the links on http://without-system.org/ ,
pro bono publico. You're welcome.


You say that is not 'a viable long-term solution for the Linux
community'? Fine, OK, whatever in Gehenna that means. But that wasn't
what I set out to do. I set out to scratch my own itch and then to give
correct, useful information to anyone with a use-case similar to my own.
Judging from feedback, some actually read the page with context and
found it useful. Others _claim_ to have read the page with context and
want to argue, but raise non-sequitur objections because they didn't pay
attention.

And of course, there are probably errors on that page. But erroneous
critiques and vague ideological appeals won't find them.


> Maybe those are still good solutions in some speficic cases, who might
> allow single individuals to continue their computing as "normal", but
> IMHO pinning won't work forever...


You might be correct, but why not?

> ...for the same reason why in just two years you already have several
> dozens basic packages depending in a way or another on systemd.


Again, what specific packages are you speaking of?

Either my documentation on
http://linuxmafia.com/faq/Debian/openrc-conversion.html about 'Which
Debian 8 "Jessie" Packages Depend on Package Systemd' (section title) is
correct or it is not. Are there packages you say are missing?

If you assert that any of the package listed there are 'basic', then I
do not agree with you about the meaning of the word 'basic', and we will
have to agree to disagree.


> That's why, although I appreciate efforts like yours and those of a
> dozen more who have been experimenting with pinning (we have done that
> as well, for months), I remain convinced that maintaining a
> systemd-free fork of a distributition, and of a fundamental one like
> Debian, is a more sustainable way of dealing with the problem. It's
> not easy, but nothing is easy :)


I agree -- of course -- than nothing is devoid of difficulties.

All I said was that, as far as I can see, lesser measures than forking
the distribution, with less effort, would have sufficed to enforce a
no-system policy on Debian-stable indefinitely. My small experiment was
a de-minimus effort without bothering to have a third-party repo of
packages (daisy-player, etc.) recompiled and repackaged without the
--with-system build option or other corrective steps. My speculation is
that such a repo would be very feasible and a successful way to
perpetuate a no-system Debian-variant community -- and as an example of
communities doing exactly that for other enforced policies applied to
Debian, cited (on the SVLUG mailing list) the Siduction and Aptosid
Debian-variant communities, successfully maintained for many years by
just a few developers as a variant on Debian-unstable, using nothing
more than third-party repos and package preferences ('pinning').