:: Re: [DNG] APT pinning no longer wor…
Startseite
Nachricht löschen
Nachricht beantworten
Autor: Daniel Reurich
Datum:  
To: Jaromil, dng
Betreff: Re: [DNG] APT pinning no longer works
On 02/12/15 23:17, Jaromil wrote:
>
> On Wed, 02 Dec 2015, Mitt Green wrote:
>
>> Good day everyone,
>>
>> As some of you have probably
>> noticed, Unstable had received
>> their updates for APT, version 1.1.
>> It had been in Experimental for long time
>> and now it landed.
>>
>> Starting from this version pinning
>> (i.e. preventing from being installed)
>> packages this way:
>> -------------------------------------------
>> Package: foo
>> Pin: release *
>> Pin-Priority: -1
>> -------------------------------------------
>> doesn't work anymore.


Actually it's just changed...

It seems to be a change in how pinning works and not a removal according
to debian/NEWS in the source:

apt (1.1~exp9) experimental; urgency=medium

A new algorithm for pinning has been implemented, it now assigns a
pin priority to a version instead of assigning a pin to a package.

This might break existing corner cases of pinning, if they use multiple
pins involving the same package name or patterns matching the same
package name, but should overall lead to pinning that actually works
as intended and documented.

-- Julian Andres Klode <jak@???> Mon, 17 Aug 2015 14:45:17 +0200

https://anonscm.debian.org/cgit/apt/apt.git/tree/debian/NEWS?h=debian/sid&id=0507225b1b437a83201d204e45b52fb9e581b354
>
> Are there alternatives for preventing packages to be installed?


Not in a friendly way like pinning does
>
> We need to decide if to incorporate this change also into Devuan.
>

Not sure, but we should see how it shakes down in unstable for a while..

> Many people will be negatively affected by yet another limitation of
> their freedom, being now unable to choose if a certain package should *not* be
> installed on their system.


May be so... but the change now allows to pin a particular package by
version rather then just a package in a release...
>
> I am wondering what is Debian's rationale to do something like that?
> anyone has more insights?


Seems to be a correction to make pinning work like it was meant to (but
never really did apparently).


>
> We are not affected: Devuan has never relied on pinning to protect our
> freedom from the systemd avalanche, but some people do and this move
> looks like the DD's left in the Debian project want to shovel their
> decisions down the throat of all users.


Indeed, today I pushed to all branches (except unstable on arm* archs) a
new "init" package that defaults to sysvinit-core, and removes
systemd-sysv as an option altogether for init. This should (I haven't
done a clean test build to confirm tho) resolve the issues with
debootstrap (even debians) trying to install systemd-sysv.
>
> ciao
>
> _______________________________________________
> Dng mailing list
> Dng@???
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
>



--
Daniel Reurich
Centurion Computer Technology (2005) Ltd.
021 797 722