:: Re: [Dng] recommendation for consid…
Página Inicial
Delete this message
Reply to this message
Autor: Luke Leighton
Data:  
Para: dng
Assunto: Re: [Dng] recommendation for consideration: keep as close to debian as possible
hellekin <hellekin <at> dyne.org> writes:

>
> On 02/14/2015 10:16 AM, Luke Kenneth Casson Leighton wrote:
> > On Sat, Feb 14, 2015 at 8:12 AM, KatolaZ <katolaz <at> freaknet.org> wrote:
> >> On Thu, Feb 12, 2015 at 04:01:58PM +0000, Luke Kenneth Casson Leighton

wrote:
> >
> >>> is it the intent of the devuan team to:
> >>>
> >>> (a) create a "fork" which will always, at all times, without fail,
> >>> require that a debian repo be placed in /etc/apt/sources.list
> >>>
> >>> or
> >>>
> >>> (b) create a "fork" of the *entire debian package repository*, such
> >>> that it will end up over time to be as completely incompatible with
> >>> debian as ubuntu is today.
> >
> *** From
> https://git.devuan.org/devuan/devuan-project/wikis/ProjectDescription
>
> "Devuan is born for a simple goal: having a systemd-free debian jessie
> to preserve freedom choice on init and decoupling between init and the
> rest of the system ... initially, it will NOT be a complete fork, but
> just a complete infrastructure to distribute a personalized version of
> debian jessie, testing and unstable where some packages from us will be
> pinned up on top of the debian repositories..."


ok, great. whilst it's close, it still doesn't clearly answer the
question, though, and i notice, also, that the above is not made clear
on the web site.

in other words, there are a lot of words on the main http://devuan.org
web site but nothing that's definitive. a lot of position statements
and aims, but nothing concrete.

to illustrate the difference, let's rewrite the above:

"Devuan has a simple goal: to make it easy for Debian users to
entirely remove systemd yet to keep their Debian systems fully
functional, up-to-date and fully compatible with Debian.

The initial method to achieve this will be to add one extra line to
sources.list (in a similar fashion to deb-multimedia), where key
strategic packages can then be replaced with systemd-free equivalents
by simply running 'apt-get upgrade'

Our immediate goal is to provide automated transition scripts integrated
into the replacement packages for debian jessie, testing and unstable,
and to maintain them indefinitely.

Whilst it may prove unavoidable, we seek to actively avoid a complete
fork of Debian (learning the lessons from Ubuntu), not least because
we wish to make it easy for users to transition between Devuan and
Debian (with and without systemd) and we appreciate that a complete
fork would make that much more challenging."


hellenkin: can you see the difference between that and what's on the
wiki (and on the web site)?

what i wrote makes the following things very clear:

 1) your debian system will not be screwed up or compromised by using
    devuan.  you will also not lose any functionality or packages.


 2) we understand the difficulty of maintaining an entire distro.
    we implicitly understand that we will not get to 1,000 maintainers
    in the immediate future, so we are being realistic and will not
    be doing a complete fork.  it's too much effort for us, and we
    recognise that you probably wouldn't trust us (i.e. wouldn't
    even want to *try* upgrading to devuan) if we created one.


 3) we're restricting the scope of what we're doing to a few key
    strategic packages, and we're going to make it easy for you to
    remove systemd.  that's our core focus.


and a few more things, besides.

if you recall, i said that my initial concern was that devuan was
giving serious consideration to including TDE and many other things
besides. i've tried TDE: it works... but because it has to replace
one of the key packages (which they haven't kept up-to-date
because they don't have enough resources) i am now in "package
dependency hell" even though i have specified to use the debian/testing
version of TDE.

this is a common problem that anyone who has regularly upgraded
a debian/testing system that uses deb-multimedia will be familiar with:
packages from two disparate repositories are *NOT* properly kept in
sync: it's simply not possible. at one point back when ffmpeg was
depending on versions 0.49 of libav and friends, my system went into
complete melt-down due to broken package dependencies. i couldn't
upgrade *anything* because of it. in the end i had to remove
deb-multimedia entirely, compile some of the packages from source (!)
to do what i needed to do, and i waited about 6 months for things to
stabilise before beginning again.

[note: with deb-multimedia i was lucky because it was not key strategic
packages, i *could* remove them. if it had been anything in the core
packages - the essential ones - i would have been *really* screwed.
and that's really the whole point of why i seek clarity on what it is
that you are doing, here].

with a small (busy) team, you are stuck in the middle between a rock
and a hard place.

on one hand you need to keep the alternative packages fully up-to-date:
even *one day* late means that people will have a system which becomes
unuseable due to version-bumps from debian.

and on the other hand you have to consider doing a complete total fork
of debian, with all that that implies (and it's one HELL of a lot of
work to replace 1,000 debian maintainers with however many people you
have), *and* you have to have decent mirror infrastructure - i think
160 Gb was the last estimate of the amount of disk space that i heard
phil was using for free.hands.com's debian mirrors (a couple years back).
and free.hands.com i believe also flat-lines a 100mbit/sec ethernet
connection at 100% capacity.

so.

can i recommend that you discuss amongst yourselves within the devuan
team as to precisely and exactly what the direction is? can we have
a consensus from the team as to whether everyone understands fully the
scope of the project, outlines a clear roadmap (and agrees to it), and
so on?

and once that is agreed can i suggest updating the web site to make
that consensus absolutely clear?

tia,

l.