1i5t5.duncan at cox
Aug 23, 2009, 7:16 AM
Post #1 of 96
(2257 views)
Permalink
|
|
Heads-up: KDEers: Particularly kde3-ers,
|
|
Those of you kde-ers, particularly kde3-ers (aka stable kde-ers),
heads-up!
If you aren't aware of the current gentoo kde (especially kde3)
situation, you *NEED* to subscribe to the gentoo-desktop list (normally
lower activity than here, so it shouldn't be a huge burden), AND check
the archives for the last couple months.
The short version: kde3 is likely going to be masked, soon, apparently
very possibly before any kde4 is ever marked stable. The current plan is
to leave kde3 in-tree but masked, probably until early next year, at
which point it'll move to an overlay, kde-sunset or similarly named,
where it'll be maintained primarily by interested users.
If any Gentoo kde3 user has the skills and time to volunteer,
particularly if you are /not/ planning to move to kde4 in the near
future, they're looking for a Gentoo kde3 dev or two, and/or skilled kde3
users who can devote time to it.
The reasoning is multi-fold. Unfortunately, while upstream KDE gets all
nasty if you try to call kde3 unmaintained and asegio famously blogged a
year or so ago that it would be maintained as long as there were users,
apparently, unmaintained is what it's becoming in actual practice,
regardless of /what/ upstream kde wants to call it. What's happening is
that they aren't strongly encouraging KDE devs to continue to maintain
the old kde3 apps. (Note that with KDE as much of FLOSS, many of the devs
are unpaid volunteers, and volunteers can hardly be forced, but strong
encouragement is certainly possible.) As a result, bugs filed on kde3
apps are increasingly being closed as unmaintained version, upgrade.
Of course, qt3 upon which kde3 depends is in similar or even worse shape
(except that it was in arguably better shape when it went unsupported, as
until then, people had been paid to keep it working, even if they'd have
otherwise preferred to be working on the newer versions), apparently not
supported any longer by its own (commercial FLOSS) upstream.
Unfortunately, all this is complicated by the state of kde4, in many ways
a mirror image of kde3 -- specifically like a mirror image in that it's
similar, but nicely reversed. kde4 is getting all sorts of developer
attention, but again despite what upstream says, it's anything /but/ as
stable and smoothly functional and polished as kde3 is. I'm normally an
early adopter, running ~arch and in fact often unmasking and even
reaching into overlays for fresh versions, often beta or rc, sometimes
even live-vcs versions direct from the upstream repositories.
Despite all that and despite the fact that upstream kde recommended 4.2
for most users and calls 4.3 fully stable, 4.2.4 was /barely/ getting
functional enough to be able to work in it well enough for me to start
transferring settings and otherwise getting serious about switching to
kde4. Despite the recommendation, in practice, as a user that regularly
runs development versions, betas and rcs, 4.2.4 was therefore /barely/
what I'd call early beta. 4.3 (as every kde4 version so far) is markedly
better than the previous version, but there's still a LOT of broken
functionality, features still rapidly evolving, etc.
kde4.3 therefore at what I'd normally consider the late-beta stage.
User's who actually used and depended on the previous version for
anything beyond basic functionality shouldn't be upgrading yet unless
they're prepared to spend HOURS, in this case, DAYS, even WEEKS,
upgrading, finding fixes and workarounds for bugs, even switching to
alternative software solutions at times when the functionality simply
isn't there. I estimate I've spent about 80 hours on the upgrade and
reconfiguring, all told. Now, a major version switch is a major version
switch, and users WILL need to spend SOME time reconfiguring and
adapting, but perhaps 20-40 hours is reasonable, NOT 80! 80 hours, two
weeks of full-time 40-hour-week equivalent work, simply indicates how
immature and broken some aspects of the project still are, thus
necessitating workarounds and the like. (If anybody wants hard examples,
I can list the issues I had and have here, but this post is long enough
without it. Ask, or check kde's general and kde-linux lists archives for
the last couple months.)
Or, put another way, there are solid reasons no kde4 is unmasked to
gentoo stable yet.
As I said, every new kde4 version is solidly improved from the previous
one, but by kde3.5, it was very very polished, very very functional, very
very fully featured, and very very depended on, at least here. kde4
/was/ basically a ground-up rewrite, and given how mature, functional and
well polished kde3 was, they had a *LOT* of ground to cover. So while
kde4 *IS* is progressing well and rapidly, it's /just/ /not/ /there/
/yet/. Rome wasn't built in a day, neither has it ever been /re/built in
a day.
I estimate that given current progress, kde4.5 will finally compare well
against 3.5. The further 4.3.x releases should be much like -rc versions
normally are, and 4.4, scheduled for early next year, should be much like
the infamous x.0 releases that early adopters that didn't hit the betas
use, but that many users forego, in favor of x.1, which should be 4.5
(scheduled for 3Q2010, minors are semi-annual and 4.3 was early this
month, so 4.5 should be ~1 year from now). Thus, 4.3 is sort of usable
for beta tester types -- requiring a lot of user workaround and
adjustment, 4.4 should hopefully be reasonably usable by ordinary people
(what kde folks claimed 4.2 was, I /do/ expect 4.4 to hit this as they /
are/ finally hitting the fit and finish bugs that make a release fit for
ordinary users), and 4.5 should finally be a mature product, nearly bug
free and usable by nearly everyone.
But kde4 is a mirror in another regard as well, as unlike most upgrades,
it seems the more advanced a user you are, the more trouble the upgrade
tends to be. This seems to be at least partially because the basic/core
functionality plus some nice eye candy was implemented first, and it was
then released, with the more advanced functionality that kde3 advanced
users depended on still broken. Thus, users who seldom change the
defaults and are easily impressed when eye candy is made the default,
/did/ in many cases find 4.2, or even earlier, usable. It's the folks
that depended on 3.5's advanced functionality that are having the worst
upgrade problems, because much of that functionality is still only
partially working.
Regardless, the fact remains that kde4.3 is not yet in a really usable
state for many, at least not without DAYS or even WEEKS worth of
workarounds, fixes, and tweaks. Of course, that makes the situation with
kde3 even more dire, as it now looks likely that Gentoo KDE users, as KDE
users on various other distributions before, will likely be rather
strongly pushed toward the immature and not yet ready new version, as the
older well functioning version goes unsupported before the smooth upgrade
path has been established. For Gentoo/KDE, that could well mean users
will find 3.x masked before any 4.x at all is keyword-unmasked to stable.
The above is further complicated by a couple Gentoo-specific factors. Of
course, being a source-based distribution, the quality of the kde3/qt3
sources affects Gentoo users (and therefore devs) more than the typical
binary distribution user. Sources that don't build without workarounds
can often be handled by the skilled binary distribution devs doing the
building for them, yet be entirely unsatisfactory for general Gentoo use
because here, every user, including those who don't know much about
upstream at all and who lack the skills necessary to do those
workarounds, has to build from source. Thus, as the upstream kde3/qt3
sources go stale and fail to build without intervention against newer
system libraries and with newer gccs, it's putting ever more strain on
the Gentoo/KDE devs and project testers to support them.
Second, it seems that no Gentoo/KDE project members are actually still
running kde3 as their normal desktop -- they've all migrated to kde4.
Thus the urgent request for skilled kde3 users, with or without an
interest in becoming a Gentoo dev, to volunteer to help out. (Still,
it's worth mentioning that apparently unlike kde upstream, there's
effective pressure, and caring devs/testers, enough to /try/ to keep it
functioning, regardless of their personal interest in it, because they
know users continue to depend on it.) How successful they are at
actually attracting such skilled kde3 users, and how long those skilled
kde3 users remain using it and how much time they have available to
invest in the project, thus /very/ much affects how long and under what
conditions Gentoo can continue to provide a usable kde3 to /it's/ users.
So where does that actually leave us?
Well, to a large extent that depends on a number of factors that remain
unknowns ATM. The current Gentoo/KDE kde4 stabilization target is 4.3.1,
which should be release in a few weeks. As I said, upstream is finally
fixing many of the remaining serious bugs, so this is reasonable, but not
assured. There's of course a couple other factors (python issues, etc)
involved whether 4.3.1 will actually make stable or not, and even if it
does, we're looking at six weeks or so, minimum (I'm not sure when 4.3.1
is scheduled for upstream release, but Gentoo policy is 30 days without
bugs, so it'd be a minimum 30 days after that). That's early October at
the earliest. If there's complications and/or it has to wait until
4.3.2, we're looking at, perhaps, stable kde4 as a Christmas present.
Gentoo's kde3 remaining time and status depends very much on the evolving
security situation, as well as how successful they are at attracting
someone, preferably someone who is or can become a Gentoo dev, to
basically dedicate themselves to it.
Apparently, upstream maintenance is in severe enough a state (again,
despite asegio's very public claim that kde3 will continue to be
supported as long as there are users, and despite the fact they get
unhappy when people say it's unsupported) that there are very real
questions about the ability to provide security updates, as the normal
stream of browser vulnerability announcements, etc, continues. Depending
on how serious a vuln is and what components are affected, etc, there's
some chance that various other distributions will continue to cooperate
in coming up with patches, but the list of distributions continuing to
ship a full kde3 is continuing to shrink. Still, there's some government
and other reasonably large long term kde3 consultancy and support
contracts in Europe, so some patches will no doubt continue to flow for
another, probably, two years anyway, regardless of mainline distribution
and upstream support.
But anyway, they're now playing it by ear in terms of security
vulnerabilities, and if a big one comes up (for all I know there may
already be one that's not yet public), and there's no forthcoming
patches, it'll mean rather short-notice kde3 masking, very possibly,
according to the summary of the last Gentoo/KDE project meeting as posted
in -desktop (the reason people concerned about kde should be following
that list, that's where those summaries go, and thus the reason I have
all this information and can post it), without a kde4 of any kind being
stable yet.
It's based on THAT that I decided to post this. People still using and
depending on kde3 **NEED** to know what could well be happening to their
desktop.
According to that summary, they do plan to keep kde3 in-tree for a few
more months, probably until early next year some time, before booting it
to the kde-sunset or whatever they decide to call it, overlay. However,
it's likely to be masked from late this year, as I said, possibly within
weeks if the security situation warrants it.
That said, if possible, they do want a stable kde4 before kde3 gets
masked -- but it's now no longer considered a given.
Meanwhile, again according to the summary, the goal before actual removal
from tree, is EITHER one of: TWO kde4 "minor" versions stabilized, OR at
least kde4.4 out, and at least ONE "minor" version stabilized. "Minor"
is in quotes, there, because it's not clear to me exactly what they mean
in that regard. "Minor" in normal usage would be 4.3, 4.4, etc, but if
that's what's intended, and a 4.3 version does indeed make it to stable,
then the two OR conditions look pretty close to identical, since 4.4
would then be the second "minor" version stabilized. Thus, I'm wondering
if they actually meant "micro" aka "patch" version, which would fulfill
the two-stable-version requirement if 4.3.1 and 4.3.2 are stabilized,
thus distinguishing it better from having a 4.4 version out and
preferably stable. Significantly, however, that's removal from the tree
to the overlay. I know I'm repeating myself but it's important to
understand, kde3 could well be masked in September, if events warrant it,
and if so, it'll almost certainly mean NO UNMASKED/STABLE KDE IN THE TREE
AT ALL for some weeks, until some version of kde4 is deemed to have
reached that level!
So as I said, currently, they plan to remove kde3 from the tree (where it
will have probably completed the last few months in-tree masked), along
with all packages depending on kde3, sometime 1H2010 (first half next
year). qt3 and all qt3 dependencies will follow shortly thereafter, so
likely before this time next year. Both will be headed to overlays, with
the viability of the overlays, at least the kde-sunset overlay, almost
certainly depending on skilled users, not kde devs.
All that can be summarized in one sentence: If you are currently a kde3
user and have NOT yet figured out where you're moving to from there, you
**BETTER** get a move on!
FWIW, they *DO* plan to announce it on the Gentoo front-page, in the user
forum, and via the gentoo tree package news mechanism, before the
masking, and likely again before the final move out of tree to the
overlay. However, given the time it took /me/ to accomplish the upgrade
and the serious trouble I had getting actually working kde4 or suitable
non-kde replacements for all the functionality I depend on, AND the usual
churn that accompanies a major desktop upgrade of that nature even if
everything technically goes off without a hitch, I decided a bit of an
additional heads-up warning would likely be appreciated by anyone still
on kde3, particularly if they've not yet started preparing for the
inevitable and now rather shortly pending.
--
Duncan - List replies preferred. No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master." Richard Stallman
|
signature.asc
1
