Re: Plasma desktop unusable in stretch

[Diederik de Haas]

On Saturday 05 September 2015 11:26:42 Martin Steigerwald wrote:
> The transition is challenging. From what I see the Debian Qt/KDE team does
> a  *marvellous* job. ...
> 
> So members of the Debian Qt/KDE team, I am now also one, work, usually in 
> their free time, unpaid, on delivering a new Plasma version to all of you.
> 
> They do a marvellous job and that in my eyes deserves a huge big "Thank
> you!"  instead of bashing on them.

I whole heartly agree.
So thanks Debian Qt/KDE team!

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

Re: Plasma desktop unusable in stretch

[Christian Hilberg]

Am Donnerstag 03 September 2015, 16:37:09 schrieb Andrey Rahmatullin:
> On Thu, Sep 03, 2015 at 12:43:37PM +0200, Christian Hilberg wrote:
> > "It's all automatic" is the bit I missed. I was under the impression
> > that this would be true for experimental->unstable only
> experimental->unstable is fully manual (even requiring a new upload).
Thanks for clarifying.


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

Re: Plasma desktop unusable in stretch

[Christian Hilberg]

Hi Brad,

thanks for clarifying matters. A few more bits inline.

Am Donnerstag 03 September 2015, 14:19:53 schrieb Brad Rogers:
> On Thu, 03 Sep 2015 12:43:37 +0200
> Christian Hilberg  wrote:
> 
> Hello Christian,
> 
> >> You appear to expect somebody (well, several 
> >> somebodies) to stem the tide to avoid problems.  
> >"Expecting" is not what I meant to express, it is more like "being
> I wasn't sure, hence my writing "appear to..."  Having cleared up a
> misunderstanding about the unstable->testing migration, I can tell
> you're a lot less angry (I use the word cautiously; it may be too strong)
> about what's going on ATM.

No anger at all. Please excuse if my writings implied this.

What you call a misunderstanding about the package migration
between unstable and testing I think is better described by
the term "misconception on my side", something many may share
with me. So thanks again for taking time to explain.

> >Sure. Since there are the brave ones running unstable, I thought this
> >is where the biggest breakages were going to be caught and eliminated
> >before the packages entered testing.
> By and large, the biggest breakages _do_ occur in unstable.  The current
> situation is, in the time I've used testing (approx ten years),
> unprecedented.  I've seen the odd package or two migrate that couldn't
> be installed, but never on such a huge scale.

That's true. I'm a debian addict myself for more than 10 years
now, and haven't seen like this either. Although, I must admit,
I have not followed testing that much close.

> [...] 
> Much of the problem, AIUI, is caused by the change to GCC5 for
> Kf5/Plasma.  Just look at the number of packages held back from testing
> because of it;
> 
> Updating gcc-5 makes 408 depending packages uninstallable on amd64: 
>
> Note that that list refers to amd64 only.  It may differ on other
> architectures.  And that's just the _depending_ packages.  There's
> another 318 non-depending packages that are affected.

It might be helpful to have some mechanism handy which would allow
the release managers to easily hold off e.g. all of KDE from migrating
from unstable into testing while something like the GCC transition
is going on, and having testing just sit on the latest pre-GCCv5 version.
Once the transition is done, this "hold" flag would be removed and
the unstable packages (for KDE here) would start trickling into testing
again. But then, for something like this to work, the package dependencies
would need to already be correct... just thinking out loud, never mind. :-)

> https://release.debian.org/migration/testing.pl?package=gcc-5 has the
> full list.  It includes stuff like LibreOffice and all sort of Python
> stuff, not just KDE.

Thanks for the pointers.

> It's going to take a long time to sort it out, I'm sure.

I might bug you with some new bug reports in the meantime. :)

Kind regards,
Christian


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

Re: Plasma desktop unusable in stretch

[Andrey Rahmatullin]

On Thu, Sep 03, 2015 at 06:30:13PM +0200, Matthias Bodenbinder wrote:
> Am 03.09.2015 um 18:12 schrieb Martin Steigerwald:
> >> Packages from Debian Unstable enter the next-stable testing distribution
> >> automatically, when a list of requirements is fulfilled: [...] The package
> >> does not introduce new release critical bugs.
> > 
> > So and now *who* reports these?
> > 
> > Then rethink from there.
> 
> Well, I would expect a minimum of testing from the package maintainers even 
> before a package is uploaded to unstable. I assume for the case at hand that 
> the package maintainers were pretty much aware about the broken KDE desktop. 
The maintainers are well aware that packages may behave badly during
large transitions and that shouldn't stop them from uploading, as it's the
only way forward.

-- 
WBR, wRAR


signature.asc
Description: Digital signature

Re: Plasma desktop unusable in stretch

[Brad Rogers]

On Thu, 03 Sep 2015 16:54:32 +0200
Christian Hilberg  wrote:

Hello Christian,

>I might bug you with some new bug reports in the meantime. :)

Not me!   :-)

Like you, I'm a Debian user.  I don't represent Debian or KDE in any
way.  All opinions are my own etc, etc.

It just so happened that I could answer some of your queries.

-- 
 Regards  _
 / )   "The blindingly obvious is
/ _)radnever immediately apparent"
I'm not here for your entertainment
U & Ur Hand - P!nk


pgpO8CqiNYCgZ.pgp
Description: OpenPGP digital signature

Re: Plasma desktop unusable in stretch

[Brad Rogers]

On Thu, 03 Sep 2015 19:06:10 +0200
Diederik de Haas  wrote:

Hello Diederik,

>allotted time, the package transitions to testing. 

I was surprised to see that "allotted time" can be as little as two
days.  Which doesn't give unstable users a lot of time to discover any
real nasties.

For most packages, transition time is ten days, as I'm sure most people
know.

-- 
 Regards  _
 / )   "The blindingly obvious is
/ _)radnever immediately apparent"
Black man got a lot of problems, but he don't mind throwing a brick
White Riot - The Clash


pgprBT5HjCy1o.pgp
Description: OpenPGP digital signature

Re: Plasma desktop unusable in stretch

[Andrey Rahmatullin]

On Thu, Sep 03, 2015 at 03:03:35PM -0400, Gary Dale wrote:
> Sorry, but before releasing a major package to "testing", shouldn't it be
> tested against the "testing" environment? 
No, we don't do this.

-- 
WBR, wRAR


signature.asc
Description: Digital signature

Re: Plasma desktop unusable in stretch

[Christian Hilberg]

Am Donnerstag 03 September 2015, 18:12:04 schrieb Martin Steigerwald:
> Am Mittwoch, 2. September 2015, 22:46:12 schrieb Matthias Bodenbinder:
> > Am 02.09.2015 um 19:50 schrieb Gary Dale:
> > [...]
> > > the brave souls who run sid do what they, not I, signed up for.
> > I 100% support that statement. On the debian wiki
> > (https://wiki.debian.org/DebianTesting) it says:
> > 
> > Packages from Debian Unstable enter the next-stable testing distribution
> > automatically, when a list of requirements is fulfilled: [...] The package
> > does not introduce new release critical bugs.
> 
> So and now *who* reports these?

If the packages should be allowed into testing only if not
introducing new RC bugs, then the bugs must be reported
(i.e., opened) by the users of unstable, if they should
prevent the package from trickling into testing.

Then again, since the packages actually /went/ into testing,
this means that there were no new RC bugs opened by the
users of unstable -- maybe the packages were working there
and only in conjunction with the package situation in testing
the issues show.

That seems to be a pretty funny puzzle to solve.

> Then rethink from there.

Trying to. =)

Regards,
Christian

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

Re: Plasma desktop unusable in stretch

[Matthias Bodenbinder]

Am 03.09.2015 um 18:12 schrieb Martin Steigerwald:
>> Packages from Debian Unstable enter the next-stable testing distribution
>> automatically, when a list of requirements is fulfilled: [...] The package
>> does not introduce new release critical bugs.
> 
> So and now *who* reports these?
> 
> Then rethink from there.

Well, I would expect a minimum of testing from the package maintainers even 
before a package is uploaded to unstable. I assume for the case at hand that 
the package maintainers were pretty much aware about the broken KDE desktop. 

Matthias

Re: Plasma desktop unusable in stretch

[Andrey Rahmatullin]

On Thu, Sep 03, 2015 at 06:16:53PM +0100, Brad Rogers wrote:
> >allotted time, the package transitions to testing. 
> I was surprised to see that "allotted time" can be as little as two
> days.  Which doesn't give unstable users a lot of time to discover any
> real nasties.
Urgencies higher than medium are used only in special cases.

> For most packages, transition time is ten days, as I'm sure most people
> know.
Five.
https://lists.debian.org/debian-devel-announce/2013/11/msg7.html


-- 
WBR, wRAR


signature.asc
Description: Digital signature

Re: Plasma desktop unusable in stretch

[Martin Steigerwald]

Am Mittwoch, 2. September 2015, 22:46:12 schrieb Matthias Bodenbinder:
> Am 02.09.2015 um 19:50 schrieb Gary Dale:
> > I know I've been using Debian for a long time. That's why I'm amazed that
> > something this bad is being put into testing. When KDE4 came along, it at
> > least was fairly stable and usable (if incomplete) before it was moved to
> > testing. If I wanted bleeding edge software that didn't necessarily work,
> > I would be running sid or pulling stuff from experimental.
> > 
> > As your quote says "Those criteria should ensure a good quality for
> > packages within testing." Putting out a broken desktop can in no way be
> > considered "good quality". I run testing because I want to test software,
> > not to get mad at the developers who volunteer their time to get the
> > packages ready for prime time. In this case the process seems to have
> > misfired by bringing us the Plasma desktop when it is nowhere near ready.
> > 
> > This is the third time in the last few weeks that I've been forced to use
> > Gnome because KDE couldn't do even basic desktop stuff. I suggest that
> > testing should revert to KDE4, if that is possible, and wait until KDE5
> > achieves some level of stability before inflicting it on us again. Let
> > the brave souls who run sid do what they, not I, signed up for.
> I 100% support that statement. On the debian wiki
> (https://wiki.debian.org/DebianTesting) it says:
> 
>   Packages from Debian Unstable enter the next-stable testing distribution
> automatically, when a list of requirements is fulfilled: [...] The package
> does not introduce new release critical bugs.

So and now *who* reports these?

Then rethink from there.

Thanks,
-- 
Martin

Re: Plasma desktop unusable in stretch

[Andrey Rahmatullin]

On Thu, Sep 03, 2015 at 12:03:04PM -0400, Gary Dale wrote:
> I've tended to view "testing" as a rolling release
That's the main misunderstanding.
We had several tries to design and/or create a rolling Debian distro but
there is nothing currently existing AFAIK.
And it takes a lot more effort than is currently spent on unstable and
testing (which is already a lot).

-- 
WBR, wRAR


signature.asc
Description: Digital signature

Re: Plasma desktop unusable in stretch

[Diederik de Haas]

On Thursday 03 September 2015 12:03:04 Gary Dale wrote:
> I also used to think of the "testing" package maintainers as the 
> guardians at the gate

There is no such thing. Package maintainers upload packages to unstable.
When there are no Release Critical bugs reported by *users* of unstable in the 
allotted time, the package transitions to testing. 


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

Re: Plasma desktop unusable in stretch

[Gary Dale]

On 03/09/15 03:42 PM, Andrey Rahmatullin wrote:

On Thu, Sep 03, 2015 at 03:03:35PM -0400, Gary Dale wrote:

Sorry, but before releasing a major package to "testing", shouldn't it be
tested against the "testing" environment?

No, we don't do this.


Obviously, but the question was "shouldn't you?"

Re: Plasma desktop unusable in stretch

[Brad Rogers]

On Thu, 03 Sep 2015 10:12:52 +0200
Christian Hilberg  wrote:

Hello Christian,

>I guess it might have been wiser to let the transitions happen in
>unstable, since the massive breakage you mention was to be expected,
>and have the smaller issues and oversights ironed out in testing.
>This scheme worked out quite well in the past.

Testing acquires packages from unstable when certain criteria are met.
It's all automatic, except for when a freeze occurs.  That is to say,
nobody oversees it.  You appear to expect somebody (well, several
somebodies) to stem the tide to avoid problems.  Question is;  How do
they know what's going to cause problems?  Answer;  They don't/can't
until the problem(s) actually arise.  By then it's too late to avoid
them.

What you'd like to happen is never going to happen.  So it comes down to
three choices;

1) deal with it yourself
2) use stable
3) use another distro

With snapshot.debian.org 1 is easy to do, 2 may require an install and
3 _will_ require an install.

1 is probably the easiest.

-- 
 Regards  _
 / )   "The blindingly obvious is
/ _)radnever immediately apparent"
It's only the children of the --- wealthy tend to be good looking
Ugly - The Stranglers


pgpQmkyld7MtA.pgp
Description: OpenPGP digital signature

Re: Plasma desktop unusable in stretch

[Christian Hilberg]

Hi Brad,

Am Mittwoch 02 September 2015, 16:16:10 schrieb Brad Rogers:
> On Wed, 02 Sep 2015 10:21:47 -0400
> Gary Dale  wrote:
> 
> Hello Gary,
> 
> >Yesterday I rebooted my computer but when it came back up and I logged 
> >in, Plasma was no longer usable.
> 
> Come on Gary, you've been on this ML long enough to know that KDE is
> going through some *massive* changes ATM.  The path from KDE4 to
> KF5/Plasma is far from an easy one to tread.  Not least because of the
> change to GCC v5.  KF5/Plasma is very, /very/ different from KDE4.  It's
> not a huge surprise, to me at any rate, that some packages don't (yet)
> have their dependencies sorted out fully.
> 
> If one finds, when doing an update, it's necessary to remove large
> numbers of packages to get everything updated then one should pause and
> consider;  Do I really want to lose half of my software suite?  Usually
> the answer is "no".  In that case, see what can be updated without
> ripping the heart out of your system.
> 
> Testing sometimes has breakage.  Sometimes that breakage is big.  You
> just have to deal with it.  If you can't
> 
> there's always stable.

To me, that kind of breakage (due to the transitions KDE4->KF5 *and*
GCC4->GCC5 at the same time) is what we're used to see in unstable.
This is what unstable is for, imho.

By letting these transitions happen simultaneously in unstable as well
as testing, the ML became flooded with all-the-same-topic mails over
and over, because many people are using testing who do so because they
like to be more recent than stable while not daring enough to expedition
into unstable land.

I guess it might have been wiser to let the transitions happen in
unstable, since the massive breakage you mention was to be expected,
and have the smaller issues and oversights ironed out in testing.
This scheme worked out quite well in the past.

Kind regards,
Christian


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

Re: Plasma desktop unusable in stretch

[anxiousmac]

On Thursday, 3 September 2015 09:20:02 UTC+1, Christian Hilberg  wrote:
> Hi Brad,
> 
> Am Mittwoch 02 September 2015, 16:16:10 schrieb Brad Rogers:
> > On Wed, 02 Sep 2015 10:21:47 -0400
> > Gary Dale  wrote:
> > 
> > Hello Gary,
> > 
> > >Yesterday I rebooted my computer but when it came back up and I logged 
> > >in, Plasma was no longer usable.
> > 
> > Come on Gary, you've been on this ML long enough to know that KDE is
> > going through some *massive* changes ATM.  The path from KDE4 to
> > KF5/Plasma is far from an easy one to tread.  Not least because of the
> > change to GCC v5.  KF5/Plasma is very, /very/ different from KDE4.  It's
> > not a huge surprise, to me at any rate, that some packages don't (yet)
> > have their dependencies sorted out fully.
> > 
> > If one finds, when doing an update, it's necessary to remove large
> > numbers of packages to get everything updated then one should pause and
> > consider;  Do I really want to lose half of my software suite?  Usually
> > the answer is "no".  In that case, see what can be updated without
> > ripping the heart out of your system.
> > 
> > Testing sometimes has breakage.  Sometimes that breakage is big.  You
> > just have to deal with it.  If you can't
> > 
> > there's always stable.
> 
> To me, that kind of breakage (due to the transitions KDE4->KF5 *and*
> GCC4->GCC5 at the same time) is what we're used to see in unstable.
> This is what unstable is for, imho.
> 
> By letting these transitions happen simultaneously in unstable as well
> as testing, the ML became flooded with all-the-same-topic mails over
> and over, because many people are using testing who do so because they
> like to be more recent than stable while not daring enough to expedition
> into unstable land.
> 
> I guess it might have been wiser to let the transitions happen in
> unstable, since the massive breakage you mention was to be expected,
> and have the smaller issues and oversights ironed out in testing.
> This scheme worked out quite well in the past.
> 
> Kind regards,
>   Christian

If we don't want breakage, we have to use stable. The primary purpose of 
testing is to develop the next release, not necessarily to produce a 
user-focused version of stable with newer packages. Last time I looked, 
warnings about this were liberally included in Debian documentation and wiki 
pages.

I have had my fair share of breakages using unstable, and trying to find my way 
out of them is usually quite educational. If I'm too busy at the time, I can 
always ssh my way to my data from another machine. Failing that, a live image.

anxiousmac

Re: Plasma desktop unusable in stretch

[Christian Hilberg]

Hi Brad,

Am Donnerstag 03 September 2015, 09:40:35 schrieb Brad Rogers:
> On Thu, 03 Sep 2015 10:12:52 +0200
> Christian Hilberg  wrote:
> 
> Hello Christian,
> 
> >I guess it might have been wiser to let the transitions happen in
> >unstable, since the massive breakage you mention was to be expected,
> >and have the smaller issues and oversights ironed out in testing.
> >This scheme worked out quite well in the past.
> 
> Testing acquires packages from unstable when certain criteria are met.
> It's all automatic, except for when a freeze occurs. That is to say,
> nobody oversees it.  

"It's all automatic" is the bit I missed. I was under the impression
that this would be true for experimental->unstable only and that the
packages in testing would then be marked as "ready for testing" manually,
once the breakage reports cease.

> You appear to expect somebody (well, several 
> somebodies) to stem the tide to avoid problems.

"Expecting" is not what I meant to express, it is more like "being
under the [wrong, I know by now] impression that somebody does",
given the actually relatively low frequency of breakages in testing.

> Question is;  How do they know what's going to cause problems?  Answer;  They 
> don't/can't
> until the problem(s) actually arise.  By then it's too late to avoid
> them.

Sure. Since there are the brave ones running unstable, I thought this
is where the biggest breakages were going to be caught and eliminated
before the packages entered testing.

> What you'd like to happen is never going to happen.  So it comes down to
> three choices;
> 
> 1) deal with it yourself
> 2) use stable
> 3) use another distro
> 
> With snapshot.debian.org 1 is easy to do, 2 may require an install and
> 3 _will_ require an install.
> 
> 1 is probably the easiest.

3) is not at all an option, sorry. ;-))

What I'm doing is using stable on my production machines and running
testing in a VM to follow up on what's going on and report issues as
I find them. It is not too much I can do there, but I try to add my
2 cent.

What had me puzzled a little was that I did not experience the KDE
packaged in testing being in the state it lately has been -- so that
is what may have misled my "expectations".

Kind regards,
Christian


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

Re: Plasma desktop unusable in stretch

[Andrey Rahmatullin]

On Thu, Sep 03, 2015 at 12:43:37PM +0200, Christian Hilberg wrote:
> "It's all automatic" is the bit I missed. I was under the impression
> that this would be true for experimental->unstable only
experimental->unstable is fully manual (even requiring a new upload).

-- 
WBR, wRAR


signature.asc
Description: Digital signature

Re: Plasma desktop unusable in stretch

[Thomas Fjellstrom]

On Thu 03 Sep 2015 04:24:40 PM Gary Dale wrote:
> On 03/09/15 03:42 PM, Andrey Rahmatullin wrote:
> > On Thu, Sep 03, 2015 at 03:03:35PM -0400, Gary Dale wrote:
> >> Sorry, but before releasing a major package to "testing", shouldn't it be
> >> tested against the "testing" environment?
> > 
> > No, we don't do this.
> 
> Obviously, but the question was "shouldn't you?"

Yeah, I had thought the point of sid was to have a place to make sure things 
weren't completely broken, then things moved to testing to finish stabilizing.

-- 
Thomas Fjellstrom
tho...@fjellstrom.ca

Re: Plasma desktop unusable in stretch

[Brad Rogers]

On Thu, 03 Sep 2015 12:43:37 +0200
Christian Hilberg  wrote:

Hello Christian,

>> You appear to expect somebody (well, several 
>> somebodies) to stem the tide to avoid problems.  
>"Expecting" is not what I meant to express, it is more like "being

I wasn't sure, hence my writing "appear to..."  Having cleared up a
misunderstanding about the unstable->testing migration, I can tell
you're a lot less angry (I use the word cautiously; it may be too strong)
about what's going on ATM.

>Sure. Since there are the brave ones running unstable, I thought this
>is where the biggest breakages were going to be caught and eliminated
>before the packages entered testing.

By and large, the biggest breakages _do_ occur in unstable.  The current
situation is, in the time I've used testing (approx ten years),
unprecedented.  I've seen the odd package or two migrate that couldn't
be installed, but never on such a huge scale.

>> 1) deal with it yourself
>> 2) use stable
>> 3) use another distro
>3) is not at all an option, sorry. ;-))

No need to apologise, it's not necessary.  In all honesty, I didn't
expect it to be.

>What had me puzzled a little was that I did not experience the KDE
>packaged in testing being in the state it lately has been -- so that
>is what may have misled my "expectations".

Much of the problem, AIUI, is caused by the change to GCC5 for
Kf5/Plasma.  Just look at the number of packages held back from testing
because of it;

Updating gcc-5 makes 408 depending packages uninstallable on amd64: 

Note that that list refers to amd64 only.  It may differ on other
architectures.  And that's just the _depending_ packages.  There's
another 318 non-depending packages that are affected.

https://release.debian.org/migration/testing.pl?package=gcc-5 has the
full list.  It includes stuff like LibreOffice and all sort of Python
stuff, not just KDE.

It's going to take a long time to sort it out, I'm sure.

-- 
 Regards  _
 / )   "The blindingly obvious is
/ _)radnever immediately apparent"
An old custom to sell your daughter
Hong Kong Garden - Siouxsie & The Banshees


pgpbMsgaNmxYU.pgp
Description: OpenPGP digital signature

Re: Re: Plasma desktop unusable in stretch

[Hillel Lubman]

On Thu, 3 Sep 2015 21:21:52 +0500 Andrey Rahmatullin wrote: 
> That's the main misunderstanding.
> We had several tries to design and/or create a rolling Debian distro but
> there is nothing currently existing AFAIK.
> And it takes a lot more effort than is currently spent on unstable and
> testing (which is already a lot).

So what happened to these efforts to make always usable / always releasable 
testing? I remember such proposals, for example:
https://lwn.net/Articles/406301/
https://lwn.net/Articles/550032/

I always viewed testing as intended at desktop usage which ideally should suit 
causal non technical users too, but at the same time I know that such goals 
weren't yet reached and things can break seriously (so using snapshots to avoid 
major issues like recent Plasma 5 breakage is something every user of testing 
should learn). That's why I'm still hesitant to recommend testing for those who 
aren't ready to deal with such issues.

So are there still any plans to change that like above, or they were stalled 
because it requires too much resources to implement and maintain?

Regards,
Hillel Lubman.

Re: Plasma desktop unusable in stretch

[Andrey Rahmatullin]

On Wed, Sep 02, 2015 at 10:21:47AM -0400, Gary Dale wrote:
> I thought testing was supposed to be where packages were mainly working and
> the sid and experimental where were packages were tested to make sure they
> actually worked. 
Not really. "The testing distribution is generated automatically by taking
packages from unstable if they satisfy certain criteria. Those criteria
should ensure a good quality for packages within testing." [0]
That doesn't guarantee the testing is usable at any single point of time,
and the packages can even be temporarily removed from testing. This only
guarantees that packages in testing are not RC-buggy and pass other
technical checks. Packages in Debian are not "tested to make sure they
actually worked" other than by the users. So, during large transitions
both unstable and testing may be temporarily unusable. Users who want
continuously working systems should use stable (or expect to make efforts
to deal with problems that happen in testing and unstable).

[0]:
https://www.debian.org/doc/manuals/developers-reference/ch04.en.html#sec-dists

-- 
WBR, wRAR


signature.asc
Description: Digital signature

Re: Plasma desktop unusable in stretch

[Gary Dale]

On 02/09/15 11:41 AM, Andrey Rahmatullin wrote:

On Wed, Sep 02, 2015 at 10:21:47AM -0400, Gary Dale wrote:

I thought testing was supposed to be where packages were mainly working and
the sid and experimental where were packages were tested to make sure they
actually worked.

Not really. "The testing distribution is generated automatically by taking
packages from unstable if they satisfy certain criteria. Those criteria
should ensure a good quality for packages within testing." [0]
That doesn't guarantee the testing is usable at any single point of time,
and the packages can even be temporarily removed from testing. This only
guarantees that packages in testing are not RC-buggy and pass other
technical checks. Packages in Debian are not "tested to make sure they
actually worked" other than by the users. So, during large transitions
both unstable and testing may be temporarily unusable. Users who want
continuously working systems should use stable (or expect to make efforts
to deal with problems that happen in testing and unstable).

[0]:
https://www.debian.org/doc/manuals/developers-reference/ch04.en.html#sec-dists

I know I've been using Debian for a long time. That's why I'm amazed 
that something this bad is being put into testing. When KDE4 came along, 
it at least was fairly stable and usable (if incomplete) before it was 
moved to testing. If I wanted bleeding edge software that didn't 
necessarily work, I would be running sid or pulling stuff from experimental.


As your quote says "Those criteria should ensure a good quality for 
packages within testing." Putting out a broken desktop can in no way be 
considered "good quality". I run testing because I want to test 
software, not to get mad at the developers who volunteer their time to 
get the packages ready for prime time. In this case the process seems to 
have misfired by bringing us the Plasma desktop when it is nowhere near 
ready.


This is the third time in the last few weeks that I've been forced to 
use Gnome because KDE couldn't do even basic desktop stuff. I suggest 
that testing should revert to KDE4, if that is possible, and wait until 
KDE5 achieves some level of stability before inflicting it on us again. 
Let the brave souls who run sid do what they, not I, signed up for.

Re: Plasma desktop unusable in stretch

[Brad Rogers]

On Wed, 02 Sep 2015 10:21:47 -0400
Gary Dale  wrote:

Hello Gary,

>Yesterday I rebooted my computer but when it came back up and I logged 
>in, Plasma was no longer usable.

Come on Gary, you've been on this ML long enough to know that KDE is
going through some *massive* changes ATM.  The path from KDE4 to
KF5/Plasma is far from an easy one to tread.  Not least because of the
change to GCC v5.  KF5/Plasma is very, /very/ different from KDE4.  It's
not a huge surprise, to me at any rate, that some packages don't (yet)
have their dependencies sorted out fully.

If one finds, when doing an update, it's necessary to remove large
numbers of packages to get everything updated then one should pause and
consider;  Do I really want to lose half of my software suite?  Usually
the answer is "no".  In that case, see what can be updated without
ripping the heart out of your system.

Testing sometimes has breakage.  Sometimes that breakage is big.  You
just have to deal with it.  If you can't


there's always stable.

-- 
 Regards  _
 / )   "The blindingly obvious is
/ _)radnever immediately apparent"
I must be hallucinating, watching angels celebrating
There Must Be An Angel (Playing With My Heart) - Eurythmics


pgpUHjkamxb61.pgp
Description: OpenPGP digital signature

Re: Plasma desktop unusable in stretch

[Matthias Bodenbinder]

Am 02.09.2015 um 19:50 schrieb Gary Dale:
> I know I've been using Debian for a long time. That's why I'm amazed that 
> something this bad is being put into testing. When KDE4 came along, it at 
> least was fairly stable and usable (if incomplete) before it was moved to 
> testing. If I wanted bleeding edge software that didn't necessarily work, I 
> would be running sid or pulling stuff from experimental.
> 
> As your quote says "Those criteria should ensure a good quality for packages 
> within testing." Putting out a broken desktop can in no way be considered 
> "good quality". I run testing because I want to test software, not to get mad 
> at the developers who volunteer their time to get the packages ready for 
> prime time. In this case the process seems to have misfired by bringing us 
> the Plasma desktop when it is nowhere near ready.
> 
> This is the third time in the last few weeks that I've been forced to use 
> Gnome because KDE couldn't do even basic desktop stuff. I suggest that 
> testing should revert to KDE4, if that is possible, and wait until KDE5 
> achieves some level of stability before inflicting it on us again. Let the 
> brave souls who run sid do what they, not I, signed up for.

I 100% support that statement. On the debian wiki 
(https://wiki.debian.org/DebianTesting) it says:

Packages from Debian Unstable enter the next-stable testing 
distribution automatically, when a list of requirements is fulfilled:   
[...] The package does not introduce new release critical bugs. 

and further
[...] so expect that new stable versions of the programs you use will 
be installed as soon as they are ready for next-stable testing

That said, by no means I would expect a completely broken KDE Desktop in 
testing. 

Matthias

Plasma desktop unusable in stretch

[Gary Dale]

Yesterday I rebooted my computer but when it came back up and I logged 
in, Plasma was no longer usable. Specifically the window borders were 
gone, including the top bar. Since the top bar contains the close window 
button, if a window didn't provide a menu bar, I'd have to bring up the 
system monitor to kill it (ironically that includes system monitor).


To make matters worse, I couldn't change the focus to other windows. For 
example, I can't type anything in iceweasel, whether in a web form or in 
the address or search bars. I can click on it which raises it to the 
top, but the keyboard input continues to go to the previous window.


I'm once again back to gnome-flashback because KDE is broken.

I thought testing was supposed to be where packages were mainly working 
and the sid and experimental where were packages were tested to make 
sure they actually worked. If KDE5 is this shaky, why aren't we waiting 
for it to become at least somewhat usable before putting it out for testing?