:: Re: [Dng] recommendation for consid…
Top Page
Delete this message
Reply to this message
Author: Luke Leighton
Date:  
To: dng
Subject: Re: [Dng] recommendation for consideration: keep as close to debian as possible
KatolaZ <katolaz <at> freaknet.org> writes:

>
> On Thu, Feb 12, 2015 at 12:46:42PM -0300, Renaud OLGIATI wrote:


> > Fork is permanent only as long as the two branches do not later converge

and rejoin.
>
> Yes, I know that, but have you ever seen anything similar happening in
> the past?


no - but that is, honestly, completely irrelevant. firstly: past
indicators have absolutely nothing to do with the future. secondly,
all of the examples you give below, as well as all of the ones that
*i* can think of, have *very good reasons* why a merge is either
(a) unnecessary (b) possible.

let's go through them, looking for possible reasons - interest on
either side - as to why *each party* would *want* to merge.


> Have Mandrake or SuSe ever reconciled with RedHat?


mandrake was a complete fork, with optimised recompiles specifically
for redhat packages. they only supported 686. also, because they
were _not_ "Redhat plc" they were required to *remove* all mention
of the word "Redhat" from all RPMs.

as that was a massive undertaking it was an irreconcilable fork.

also, what possible interest would redhat have to reconcile what
mandrake had done?

no reason that i can think of.


suse, likewise, was an irreconcilable fork due to name changes, and
a complete direction change.

again, there is no possible reason that i can think of as to why suse
would want to merge with redhat, not why redhat would want to merge
with suse.


> Will Ubuntu or Mint or GnewSense ever merge again with Debian?


ubuntu was a fuck-up that annoyed a hell of a lot of people. canonical
decided that the best way forward was to make a complete and total fork
of all of the debian packages, change the release organisation, change
the release management, change even the damn package names.

so not only is a merge flat-out impractical for _technical_ reasons, but
debian developers are generally quite happy that ubuntu exists because,
as ubuntu is designed to be idiot-proof, it keeps the idiots away from
the debian mailing lists.

assessment of the possibility and *desire* of a debian-ubuntu merge: ZERO.

mint: that's an interesting one. i haven't looked closely but my
understanding is that it's an option (a) style fork [requires two
repos, one which replaces key packages of the other but no more].
if that's the case, then one would be left with "what desire is
there for the two to merge", and that, obviously, requires canonical
to consider adding the mint developers to the payroll of canonical.
this might not be something that they wish to do.

GnewSense: GnewSense is an interesting one because it has stricter
software libre package selection criteria even than debian. not just
the "non-free" repo is removed but much more. also they will have
name-changes (replacements of incorrect occurrences of "Linux"
where it should be "GNU/Linux" and much more to consider). all that
makes it less likely for a merge to be considered... but not completely
impractical.

which leaves "desire" to merge to be considered. what *desire* is there
for GnewSense to be merged into debian? well, if GnewSense is merged
into debian, it does so by destroying the entire very founding principle
of GnewSense! why is that so? because to be merged into Debian, it
becomes possible - easily possible - for end-users to either deliberately,
accidentally or unknowingly add in non-free packages by editing
sources.list.

so there is *no way* that the GnewSense developers would even *want*
GnewSense to be merged with Debian, as it is founded on much stronger
Software Freedom Principles.

then there is deb-multimedia, which, similar to GnewSense, goes the
*opposite* way. there is therefore, likewise, absolutely no *desire*
to merge deb-multimedia into debian because that would be in violation
of the *debian* charter due to the patent and licensing issues around
the deb-multimedia packages.


does this give enough of a strong illustration as to why your statement
is not applicable, KatolatZ?

let's look then at debian-devuan. firstly it *is* possible to
create a re-mergeable fork, if care is taken over the package
creation, and the scope of devuan is kept strictly under control.

and secondly this is one of the very rare circumstances where i
believe it would be *desirable* for a future merge to actually
take place. i mentioned this in my previous post, that devuan
could be a "testing ground" for radical ideas such as shredding
systemd, which debian could not possibly do without a massive
amount of risky disruption.

> Nope. I'm
> sorry but merging does not in fact exists for distributions.


i trust that i have illustrated that this is a short-sighted
self-fulfilling conclusion to reach?

> > Who knows, maybe the Debian devs will realize they are missing

something, and integrate the non-systemd
> in a later release...


how will they do that if you have made it impossible - technically -
for them to consider doing that?

if you set a strict rule of keeping to a strict subset of core
packages, use "Provides" or other technique to ensure that the
packages are direct one-for-one replacements, then there stands
a chance that what you are working on may be proposed for inclusion
into debian.

if on the other hand you "Do An Ubuntu" there is *guaranteed*
absolutely zero chance.

and, to be honest, i will not be interested in supporting you
or using devuan if it becomes "Another Ubuntu". i've seen what
ubuntu has become, and i won't use it - ever.


> Oh yes, it might be. But it is very unlikely, since in two years it
> could basically mean maintaining two debians...
>
> I am sorry but revolutions are not cheap, and every time you choose to
> go for something then you have to give up on something else. I really
> hope Debian will reconsider the systemd nonsense, but I suspect that
> the probability for this event to occur is practically zero. Hence, I
> must conclude that Debian and Devuan will most probably not merge any
> time soon, and will most probably keep diverging instead, as it has


this is again a self-fulfilling statement of intent, where i have
demonstrated logically and rationally above that the grounds for the
conclusion that you draw are incorrect.

is there anything that you can perceive which is incorrect about the
rationale i give above?

l.