Kubuntu LTS

[Luka Renko]

On Monday 24 December 2007 11:43:59 Scott James Remnant wrote:
> Hi Luka,
>
> The Technical Board was not formally involved[0] in the decision whether
> or not to release Kubuntu 8.04 as an LTS.  "LTS" is a commercial support
> commitment provided by Canonical Ltd, who shoulder the financial and
> administrative burden of doing so; as such, it is entirely their
> decision as to whether or not they provide that support for a particular
> release.
>
> They make the decision on a purely commercial basis, determined by
> factors such as: the stability of the release in question; the
> supportability of the release; the commercial interest in the release,
> etc.
>
> Given that, and since the remainder are not technical in nature or
> resolving a technical dispute, the Technical Board cannot answer your
> questions.

Thank you for making this very clear. In early discussion, Technical Board was 
mentioned a lot (as the authority to explain the decision) and this is why I 
have put it on Cc for clarification.


> However since I was involved in the decision-making process at Canonical
> as part of my job as leader of the Ubuntu Desktop team, I will attempt
> to answer your questions as best I can.  The rest of this e-mail shall
> not be considered a reply from the Technical Board.
>
>
> Kubuntu 8.04 comes at a difficult time in the KDE release cycle, it
> arrives just after the dawn of a major new release of the platform which
> is a very significant new amount of work.  This is a new platform that
> is receiving a significant amount of attention, both from its developers
> and from its users.

I completely understand KDE4 brings new momentum and therefore needs to be 
taken into the picture. My concern is that such thinking and decision is made 
that late in Hardy schedule (even more being LTS targeted release). Everybody 
kind of expected that situation was clear after UDS in Boston (where all 
blueprints mention KDE 3.5 being the default also due to LTS badge).
Since KDE4 position did not change much after UDS, it is even less clear why 
the approach was changed later.


> Questions that were asked:
>
> Will KDE 4 be stable enough to support for the term of the release?
>   I've not seen anybody who believes that this would be the case; a
> long-term supported release would have to be based on the stable KDE 3.5
> series.  

Agreed and also supported by Hardy blueprints as defined on UDS.


> This gives us the second question: 
>
> Will a bug in KDE 3.5 receive upstream attention in March 2011?
>   In order for Canonical to make a commercial commitment to their
> customers, who have signed contracts and terms of service with them,
> they need to be sure that they can honour the terms of their agreements.
>   The KDE upstream position appears clear, KDE 4 is the focus of
> developer attention; KDE 3.5 will be supported as long as KDE 4 isn't
> suitable for support.
>   Given the attention being paid to KDE 4, it is difficult to believe
> that this will not be the preferred release in three years time.  Thus
> it is difficult to guarantee that upstream will still support the
> current stable series for the timescale required.

Canonical for sure can best evaluate their existing customers that are paying 
for support of Kubuntu and need to ensure that they can follow terms of 
service.
On the other hand, I do not see where the claims that KDE 3.5 would be 
completely unsupported very soon (by upstream and other actively supporting 
it today). We have to be aware that there are other commercial offerings 
based on KDE 3.5 and at least one response [1] makes it very clear that there 
are other companies besides Canonical that will have to support KDE 3.5 until 
2011 (extended support even until 2013) - not to mention that this particular 
company have several payed KDE developers on staff that supports such claims. 
Taking into account that KDE 4.0 is not even feature-wise on par with 3.5 
(and we all hope this will be not as big issue with 4.1) and that it is to 
some extend revolutionary (and not just evolutionary) upgrade, it is expected 
that more conservative customers would appreciate having stable, supported 
KDE 3.5 distribution based on Debian.


> Is the development community able to sufficiently test the new release?
>   Again, in order for the commercial commitment to be made, Canonical
> has to believe that the development community (which includes its own
> staff) are able to provide sufficient testing of both the new release
> and upgrades between older releases and the newer one.
>   This judgement is both based on the number of active testers for the
> previous release according to the ISO test tracker, and the relative
> popularity of the available packages/seeds.
>   The number of Kubuntu developers testing releases and recording their
> results is right now at an all-time low, this makes it difficult to
> guarantee sufficient test coverage of installs and most importantly
> upgrades.

It is clear that Canonical has currently one full-time KDE developer on staff 
(even though lots of other Ubuntu developers contribute a lot also for 
Kubuntu features) and I can understand it has to prioritize how to use the 
limited funds. I would also personally like to thank Canonical for all 
resources that have bin put in past years into Kubuntu.

But you have uncovered another point which seems to have influenced the 
decision: community involvement in testing releases. I was completly unaware 
that this has become an issue (myself being active in Ubiquity/ISO testing in 
past releases), even less being seen as major concern for Kubuntu in general. 
If this is seen as showstopper for proper Kubuntu support by Canonical, it 
would be good if these concerns would be communicated to Kubuntu community. 
Since we have Kubuntu Community Council, it looks like the right forum to 
address such issues.


>   Likewise the current download interest is entirely directed towards
> KDE 4 packages and CD images.

That is good to know, but we have to be aware that KDE4 is hot topic and lots 
of people download it to preview new technology and that LTS is anyway 
targeting very different customer base. This have been seen also in this 
thread on the mailing list and blogosphere, with several responses by users 
(and not just Kubuntu developers) that they have cleary choosen LTS for good 
reasons (they cannot cope with 6-month upgrade cycles...)


> I hope that provides some insight as to how this decision was made,
> expanding on the statements that Jonathan made in his announcement.

It does and I would really like to thank you again for discussing it with the 
community. 

> As to your questions about the timing, method and discussion for the
> decision:
>
> The community's input was actually sought on several points, and many
> members of the Kubuntu community provided answers and insight that
> contributed to the decision.  It is difficult for this decision to be
> made by the community because the community's stake in Kubuntu is one of
> personal achievement and pride, whereas Canonical's is financial and of
> commercial commitments.  Had Canonical simply asked the community
> "should Kubuntu 8.04 be an LTS?", the answer would not be based on the
> same terms: instead more direct questions were asked such as "how long
> will upstream work on KDE 3.5?"

I am fine with the fact that it is up to Canonical to make the decision about 
what level of commercial support it will provide. Concerns of the Kubuntu 
community are only with the fact that decision might have been made on 
questionable base and with wider impact on Kubuntu positioning in general.


> The timing of the decision was simply based on when it was clear from
> the KDE release schedule that it was inevitable that KDE 4 would be
> released before Kubuntu 8.04.  Since the KDE upstream release schedule
> is not in alignment with Ubuntu's own, it is difficult to align such
> decisions to it.

Again, situation regarding KDE4 did not change that much from October (when 
UDS was planned) until end of the year (when decision was made) that could 
explain such change in strategy regarding Kubuntu LTS.
Even though I am a supporter of fixed schedules myself, I can understand why 
such approach is even less appropriate for KDE 4.0 (being major platform 
upgrade/change) - I still hope future KDE 4.x releases would be more 
predictable. On the other hand, I never had a feeling in past that Kubuntu 
developers were not able to get latest&greatest stable KDE release into 
Kubuntu (and I am using Kubuntu from first Hoary release and contributing 
since Breezy).


> As to the unclear reasoning behind the decision, there is insufficient
> argument in your e-mail to answer that since you don't discuss where you
> feel the communication is unclear.  The announcement states that the
> reason is that KDE 4 will be released before Kubuntu 8.04, that it is
> the focus of user and developer attention, and that this upheaval is
> contrary to Canonical's requirements for an LTS.  This is the reasoning.

Again, nobody thinks that KDE 4.0 should be considered for LTS, I would 
personally leave it in universe for Hardy, allowing more people to contribute 
to it. KDE 3.5 is very stable and will not receive much new features, but it 
is still receiveing bug fixes (you can check in SVN) and I am sure at least 
3.5.9 is around the corner (even candidate for Hardy). By definition, I would 
expect this would make it even more suitable for LTS.

I have another concrete question for Canonical: will Kubuntu Hardy be still 
available through ShipIt? Will ShipIt support only KDE 4.0 or KDE 3.5 or 
both? As you have probably seen in Planet Ubuntu, there are already different 
stories regarding this topic (ranging from KDE 3.5-only, KDE 4.0-only and 
providing option for user to choose from) and this is causing quite some 
confusion.


> Your other questions only have answers in further questions:
>
> Kubuntu will be suitable for an LTS when the KDE packages it contains
> are expected to be stable, maintained and supported for three years
> afterwards.  (This could be considered LTS requirement zero; 6.06 came
> at a perfect time, it is hopeful that KDE 3.5 will at least have some
> remaining support and maintenance for the next 18 months while it
> remains supported.)

Skipping one LTS release for Kubuntu has wider impact than just not having 
3-year support for specific release:

1. Introduces uncertanty for existing Kubuntu users (and not just LTS users)
Will I need to upgrade every 6-month now? What support (from Canonical) can be 
expected in future? Will there be Kubuntu LTS in future? Can LTS be expected 
with next release (hardy+1, 8.10?)?

2. Makes even more differentiation between Ubuntu/GNOME and Kubuntu/KDE
It is clear that priorities were set much before Kubuntu was introduced with 
Hoary, but such decisions just strenghten the view that Canonical/Ubuntu does 
not really care about Kubuntu (see [1] for example).

3. Sends bad message to KDE developers (and Kubuntu community)
Canonical does not trust KDE 3.5 as being stable/supportable for 3-years (even 
though past version were considered as good enough). Kubuntu is used by lot's 
of KDE developers and good KDE4 packages just strangthen this position. 
Dropping support level for stable 3.5 release does not help in this.

I am not sure if above concerns were part of your decision making, but for 
sure that they should be. And I think it should be taken into account when 
Canonical will communicate LTS decision to wider public.

 
> No claim or commitment has been made to how LTS upgrades will work; the
> intent is that one LTS is directly upgradable to the next, however this
> requires extensive testing on behalf of the development community.

We all understand upgrade issues are big even between 6-month releases, not to 
mention 2-year releases as it is currently the case with LTS. Skipping one 
LTS for Kubuntu just increases the probability that such upgrade would be 
very hardly achievable, as even the common parts (shared with Ubuntu) would 
need lot's of special treatment which I really doubt it will make sense for 
Canonical to do just to cover Kubuntu LTS -> LTS upgrade.


> I cannot answer KDE 3.5 to KDE 4.x upgrade questions, those are
> technical.

Yes this is separate story and will for sure require nice blueprint on next 
UDS. ;-)

I would also like to stress here that it would be good if Canonical would make 
official statement for Kubuntu users regarding LTS decision: currently all 
information is discussed on IRC, kubuntu-devel mailing list and blogosphere 
and I am not sure that different personal statements with interpretations by 
Kubuntu community members are helping in clarifying the position.


Thanks again and I wish you and your family all best in 2008,

Luka

[1] http://www.kdedevelopers.org/node/3167