[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: KDE SC 4.10



Hello,

On Friday 05 April 2013 08:42:54 Sune Vuorela wrote:
> On 2013-04-05, Martin Steigerwald <Martin@lichtvoll.de> wrote:
> > What was the reason for qt-kde.debian.net then? From the page:
> Some people thinks experimental is hard. Other people dislike doing the
> copyright-documentation that is required for the official archive. And
> qt-kde.d.n also could allow people who isn't a DD/DM to put packages
> there.

Well, qt-kde.d.n as distribution channel has probably outlived its purpose 
even if it can still be useful sometimes.

The rest of the mail will be a bit OT for this list, however, in my opinion, 
users might still be interested what some of the challenges are with KDE 
packaging. Actually, these are the main things which turn (have turned) me off 
from KDE packaging these days.

IMO, it is very complicated to maintain anything that is more like 5-10 highly 
coupled source packages in Debian. You have to spend so much time on internal 
development infrastructure (constantly) that little time (or motivation) 
remains to do actual packaging changes. And as far as I know, KDE approaches 
100 source packages, so do the math (funtunately, the number of core packages 
is low). Especially, it is very expensive (in terms of both time and knowledge 
required) to start KDE packaging for the first time or "resume" work after 
longer time of inactivity.

I wish there was some "Continuous integration" for KDE packaging which took 
the load of:

* Package building and dependency management.
* Package uploading to development repository for testing.
* Automatted Lintian reports and other Q/A.
* Any other repetitive, boring but useful tasks.

experimental is not suitable for that because it's a distribution channel 
rather than a development one. Nowadays experimental latency is OK for 
distribution channel however it is very slow for development (anything more 
than 5-30 minutes is slow because time is expensive). Not to mention the fact 
that you don't want to distribute half baked packages for the sake of yourself 
and your users since you still need to test before distribution.

Even for distribution alone (no testing), uploading 100 source packages needs 
way too much of mantime. Once you do this a couple of times, it is no fun at 
all.

P.S. It may sound that monolithic KDE packages were better. Well, no. They 
were awful with respect to actual packaging tasks (where is all the fun) 
however "development infrastructure" was more manageable with fewer source 
packages (even if their size was a huge disadvantage and PITA).

Attachment: signature.asc
Description: This is a digitally signed message part.


Reply to: