About basic rules of package maintenance in Debian

[Mike Gabriel]

Hi Adrian, hi all,

On Mi 14 Aug 2013 06:42:45 CEST John Paul Adrian Glaubitz wrote:

> For some reason, Mike Gabriel has now uploaded development versions of
> both mate-common and libmatekbd into unstable. There was no announcement
> and no coordination whatsoever. He just bumped these packages
> to a completely new upstream version, ignoring the fact that this
> will break MATE for all people which are using 1.6.x from the
> MATE Debian repositories and the fact that he is introducing a
> development version of MATE into Debian unstable which is completely
> undesired as explained before.

I have to deeply apologize (publicly) for my over-enthusiastic action  
two nights ago. I was not aware of the odd/even number scheme in MATE  
and I should surely have communicated. This will not happen again!

I have communicated with Adrian yesterday quite a lot (Cc: Stefano on  
the mail communication, some communication also was on IRC) and he is  
(after being quited pi...) kind enough to help in pulling things  
straight. Thanks Adrian for your help on this!!!

> When Stefano and I started working on packaging MATE for Debian, we
> agreed on uploading only stable releases to Debian unstable which
> is common practice for all other major desktops used in Debian
> (KDE, GNOME, XFCE). The reason is that both Debian and any of
> the desktop environments have long release cycles and it is very
> difficult to coordinate the availability of the latest stable packages
> of a desktop in Debian when we're about to freeze.

> Since desktops consist of a large number of packages which take
> some time to be packaged and migrated into testing, it is very
> important to be have always a consistent and stable set of packages
> in testing such that when the freeze happens in Debian, the set
> of desktop packages is ready to be shipped in stable beside some
> minor bugs. Once we're frozen, it is NOT possible to upload new
> upstream versions.
>
> MATE and many other desktop environments use odd version numbers
> to indicate that the particular package is a development version
> which is the case for 1.7.0 as well.

Thanks for explaining this in depth. I have moved this and other bits  
of information from your mail to the (new) Wiki page(s) of the MATE  
packaging project [1].

> We are now facing the situation that we basically have beta versions
> of MATE in Debian unstable which we need to get rid of. Since it
> is not trivially easy to remove packages from the Debian archives,
> we will have to manipulate version numbers in order to fix the
> issue. For that matter, there are two possibilities.
>
> One of them are epochs, the other ones are version numbers of the
> scheme 1.7.1.really.1.6.1-1 which will overwrite the 1.7.x
> versions.

After the communication with Adrian yesterday, the way he addresses now is:

   o block migration of my faulty uploads to Debian testing
   o request removal from the FTP masters of the 1.7.x packages I upload
   o re-upload to Debian via a repeated NEW review

Correct me, Adria, if I missed anything or got anything wrong.

For me this has been quite a learning process. Thanks again to Adrian  
for jumping in and giving his time to this.

Greets+Thanks+DeeplySorry,
Mike

[1] https://wiki.debian.org/PkgMate/UpstreamReleases
[2] https://wiki.debian.org/PkgMate/StablePackaging

-- 

DAS-NETZWERKTEAM
mike gabriel, herweg 7, 24357 fleckeby
fon: +49 (1520) 1976 148

GnuPG Key ID 0x25771B31
mail: mike.gabriel at das-netzwerkteam.de, http://das-netzwerkteam.de

freeBusy:
https://mail.das-netzwerkteam.de/freebusy/m.gabriel%40das-netzwerkteam.de.xfb
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digitale PGP-Unterschrift
URL: <http://lists.alioth.debian.org/pipermail/pkg-mate-team/attachments/20130815/4948cfef/attachment.sig>