Update: We have decided going for KDE 3.5.9 in Lenny.
I would love hearing opinions, specially from stable users about what they expect/want in stable.
Below, follows an edited version of my email with my personal opinion.
My proposal is shipping KDE 3.5.9 with the KDE 4.1 development platform: kde4libs, kdepimlibs and kdebase-runtime.
No huge advantages in shipping KDE 4.1
Honestly, I do not see any advantage in shipping KDE 4.1 instead of KDE 3.5.9 besides of coolness and bleeding-edge stuff. I do see this movement more like a change of desktop than an improved new version of KDE. The KDE 4 series clearly represent a big change in innovation and improvement over KDE 3, something that has just started. KDE 4 will have a lot of to tell in the future (4.2, 4.3..), you only have to see how much it has improved from 4.0 to 4.1 beta 1. But I do not see KDE 4.1 still fully replacing all the necessities of our users. There is still a bunch of small details there and here, I will talk of some here further on this mail (for example, koffice), but what worries me here specially is that some are totally unknown for us, since we have a lot of users with very different use cases.
KDE 4.1 has not been released yet.
KDE 4.1 has not been released yet, looking at the release
it is supposed to be released July 29th, this is already impossible with
current Lenny's release, and we would need an exception from the release
team that will be only granted if it is really worth it. Then, a delay
in this schedule from KDE release team would be bad for us, since we are
so tight in time.
The sooner we could upload something to unstable would be with the RC1 that will be released on the 15th of July, 2 weeks before the final version. Lenny is supposed to go into full freeze in the mid of July, this could be delayed, but what is sure libraries will be frozen in 3-4 weeks, and we need ship a huge amount of new libraries. Besides, it is usually better ship an update of 4.1.x that contains fixes to the most important problems found in the 4.1.0 release.
Build dependencies we need to take care of
And then, it is not only the KDE 4 desktop, there are a set of build dependencies we have to maintain and not all of them are already in unstable, those dependencies are: akonadi, automoc (already in unstable, but it is a snapshot), decibel, soprano, tapioca-qt and telepathy-qt.
Some widely used apps of the KDE desktop are not ready
Even if they are not shipped with the core packages, some apps belong to the KDE desktop and to the KDE project. We have here Koffice, kdevelop or amarok. Amarok is one of a lot of widely used apps in the KDE desktop that won't have a substitute in KDE 4. There is a myriad of small apps in this situation.
Koffice 2 won't be ready so we are shipping with koffice 1 that needs
some parts of KDE 3 to work properly (like kcontrol), so it will need
some hacking because it is not currently fully installable in KDE 4...
and it won't be properly integrated anyway. I'm not sure how well works
kdevelop with KDE 4, but newer kdevelop (that needs now kdevplatform
package) won't be ready.
Quanta needs as well kdevplatform, but quanta is shipped inside kdewebdev and it is one of the modules we have not packaged yet (together with accessibility and bindings).
No positive feedback about shipping it
My blog post about how to install KDE 4.1 beta 1 had some feedback of
other developers in planet and specially from users in several blogs and
forums that linked my post. Most of the people seem to like it as in "it
will be cool", but everybody seems to agree that it is not still a
replacement for KDE 3 in their daily tasks.
Basically, I see 3 kinds of users: developers, power users and average users. KDE 4.1 could be ready for developers who are able to cope with the lack of some apps, mixing kde3 and kde4 without risks etc, and the same goes to power users. But I do not see it ready for final users, they won't like this (imposed) change.
We would be shipping the KDE 4.1 desktop just released, and usually in the community, average users wait until more advanced users are used to this new software then it is when they adopt it, because then, they have support and help from forum and mailing lists from this more advanced users.
With the current beta power users and developers still do not see it ready for daily use, so what you can expect for our average users.
Some arches do not like KDE 4
Let's keep this short: KDE 4 needs to be built in all the release archs, and it actually does not.
You can have everybody happy
My proposal is release with "old" reliable KDE 3.5.9 not forcing anyone
to update to KDE 4.1, and just provide 4.1.x "official" backports. So
whose who want to use KDE 4.1 will just use the backports and besides,
these users will have these backports updated through the KDE 4.1.x
(x=1,2,3..) updates. I am willing to work on these backports.
By the way, Backports infrastructure is only for stuff in testing, but I'm sure we can find a nice solution here.