Re: Migrating Windows CI to Qt 6.7 - Roadblocks

[Ralf Habacker]

Am 26.05.24 um 11:42 schrieb christ...@cullmann.io:


For the re-use of instances you need to use something like

https://github.com/KDAB/KDSingleApplication

or similar.

Any implicit starting of the dbus stuff will often just result in hangs
or other misbehavior.


DBus on Windows starts automatically when a client requests a dbus 
connection, so there is only a problem if dbus-daemon and the required 
configuration files were blacklisted during packaging and are not 
included in the installation. I have never had problems with this when 
working with KDE applications using MinGW.




It is just like X11: don't use it on systems that don't have it as
native windowing system, we guard that the same way.

There is a big difference to X11. X11 is usually not distributed with 
KDE applications as it can be with dbus, as there is a corresponding 
package in Craft.


There is e.g. the remote control area which is implemented in KDE 
applications with dbus. Should we now use a separate implementation for 
each platform with additional effort, although there is a 
platform-independent solution ?


Just my 2ct.

Re: KDE Gear projects with failing CI (release/24.02) (2 April 2024)

[Ralf Habacker]

Am 10.04.24 um 10:51 schrieb Ben Cooksley:

umbrello - 3rd week
  * https://invent.kde.org/sdk/umbrello/-/pipelines/657665 


   * craft_windows_qt515_x86_64 fails



This issue depends on a special dependency requirement in a third party 
package, see 
https://invent.kde.org/packaging/craft-blueprints-kde/-/blob/master/extragear/kdevelop/kdevelop/kdevelop.py?ref_type=heads#L30


Currently I have no idea how to fix this - Any hint welcome.

Regards
Ralf

KDE CI SignTool Error: No certificates were found that met all the given criteria.

[Ralf Habacker]

Hi

building kmymoney releases on KDE CI are currently broken by an
installation issue
https://binary-factory.kde.org/view/Windows%2064-bit/job/KMyMoney_Release_win64/1081/console.
Please fix.

Regards
Ralf

Re: Fwd: KDE CI: Administration » Dependency Build Applications stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 52 - Still Failing!

[Ralf Habacker]

Am 12.07.21 um 09:35 schrieb Ralf Habacker:
> 
>> The fact that you have to apply additional overrides here indicates to
>> me that your changes to 'kdewin' are also incorrect.
> 
> It may also be caused by the kdelibs4support package, as I initial
> stated - I double checked in kdewin git repo, that it exports the
> correct header and library functions.
> 

At https://build.kde.org/job/Administration/job/Dependency Build
Extragear stable-kf5-qt5 .15/55/consoleFull you can see

13:57:34  -- Looking for inet_pton
13:57:34  -- Looking for inet_pton - not found
13:57:34  -- Looking for inet_ntop
13:57:34  -- Looking for inet_ntop - not found

that  cmake does not find inet_ ... function, which *are* available by
msvc compiler since Windows Vista
(https://docs.microsoft.com/en-us/windows/win32/api/ws2tcpip/nf-ws2tcpip-inet_pton)


I guess using the same approach as done now in kdewin at
https://invent.kde.org/packaging/kdewin/-/merge_requests/1/diffs?commit_id=e7ba87c10528b71ec53865c661dd718447494591#a9f1c36d35a73bf675ad0d1641cad7d3df55284e_27_28

should fix that issue. My earlier mentioned comment related to add the
-DHAVE_INET_PTON and -DHAVE_INET_NTOP is a valid workaround.

Regards
Ralf
>>
>>>
>>> -- Forwarded message -
>>> From: *CI System* mailto:nore...@kde.org>>
>>> Date: Sat, Jul 10, 2021 at 7:21 AM
>>> Subject: KDE CI: Administration » Dependency Build Applications
>>> stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 52 - Still Failing!
>>> To: mailto:bcooks...@kde.org>>
>>>
>>>
>>> *BUILD FAILURE*
>>> Build URL
>>> 
>>> https://build.kde.org/job/Administration/job/Dependency%20Build%20Applications%20stable-kf5-qt5%20WindowsMSVCQt5.15/52/
>>> 
>>> <https://build.kde.org/job/Administration/job/Dependency%20Build%20Applications%20stable-kf5-qt5%20WindowsMSVCQt5.15/52/>
>>>
>>> Project:Dependency Build Applications stable-kf5-qt5
>>> WindowsMSVCQt5.15
>>> Date of build:  Fri, 09 Jul 2021 18:58:42 +
>>> Build duration: 22 min and counting
>>>
>>>
>>> *CONSOLE OUTPUT *
>>> [...truncated 51900 lines...]
>>> [2021-07-09T19:20:03.112Z]
>>> 
>>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>>> warning C4005: 'Q_OS_WIN': macro redefinition
>>> [2021-07-09T19:20:03.112Z] ..\src\kdeui\kicon.cpp: note: see
>>> previous definition of 'Q_OS_WIN'
>>> [2021-07-09T19:20:03.112Z] [153/533] Building CXX object
>>> src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kdialogqueue.cpp.obj
>>> [2021-07-09T19:20:03.112Z]
>>> 
>>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>>> warning C4005: 'Q_OS_WIN': macro redefinition
>>> [2021-07-09T19:20:03.112Z] ..\src\kdeui\kdialogqueue.cpp: note:
>>> see previous definition of 'Q_OS_WIN'
>>> [2021-07-09T19:20:03.112Z] [154/533] Building CXX object
>>> src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\ktzfiletimezone.cpp.obj
>>> [2021-07-09T19:20:03.112Z]
>>> 
>>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>>> warning C4005: 'Q_OS_WIN': macro redefinition
>>> [2021-07-09T19:20:03.112Z] ..\src\kdecore\ktzfiletimezone.cpp:
>>> note: see previous definition of 'Q_OS_WIN'
>>> [2021-07-09T19:20:03.112Z] [155/533] Building CXX object
>>> src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kapplication.cpp.obj
>>> [2021-07-09T19:20:03.112Z]
>>> 
>>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>>> warning C4005: 'Q_OS_WIN': macro redefinition
>>> [2021-07-09T19:20:03.112Z] ..\src\kdeui\kapplication.cpp: note:
>>> see previous definition of 'Q_OS_WIN'
>>> [2021-07-09T19:20:03.684Z] [156/533] Building CXX object
>>> src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kstringvalidator.cpp.obj
>>> [2021-07-09T19:20:03.684Z]
>>> 
>>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>>> warning C4005: 'Q_OS_WIN': macro redefinition
>>> [2021-07-09T19:20:03.684Z] ..\src\kdeui\kstringvalidator.cpp:
>>> note: see previous definition of 'Q_OS_WIN'
>>> [2021-07-09T19:20:03.684Z] [157/533] Building CXX object
>>> src\CMakeFiles\KF5KDELibs4Supp

Re: Fwd: KDE CI: Administration » Dependency Build Applications stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 52 - Still Failing!

[Ralf Habacker]

Am 10.07.21 um 21:00 schrieb Ben Cooksley:
> On Sat, Jul 10, 2021 at 9:14 PM Ralf Habacker  <mailto:ralf.habac...@freenet.de>> wrote:
> 
> Am 09.07.21 um 21:26 schrieb Ben Cooksley:
>> Hi Ralf,
>>
>> I'm afraid your previous fixes are insufficient - we are still
>> failing with a linking error.
>>
>> The following is notice that I intend to back all changes out in
>> kdewin and restore the repository to
>> 8503ac6e0e07099c4938cbfe60e3f5e25e2ec368 to restore the correct
>> function of the CI system in 24 hours time.
>>
> Please don't that not - that fix is required for building
> kdelibs4support with mingw headers version 9 and only export
> replacement functions not provided by the os. 
> 
> The platform you are talking about (mingw cross compilation) is not one
> generally supported by KDE on Windows - I believe you are the only
> person who uses that approach.
> 
I'm sorry to tell you, that your informations seems to be outdated.
There are several linux distributions out there already providing cross
compiled KDE package.


> The MingW version we do support (native compilation), 

This is simply based by the fact, the they probably do not use mingw 9.x
headers.

> along with MSVC (as the principal compiler) do not have the issue you are 
> describing.

That there is now an issue with msvc looks to me as that it is caused by
the mentioned fact, that detection for inet function seems to be broken
in kdelibs4support package, now as the correct way has been added to the
kdewin package.

If you think that it is broken in kdewin package, please show me where
it is wrong at https://invent.kde.org/packaging/kdewin/-/merge_requests/1.


> This is how all the Binary Factory builds are run, and therefore how
> every project on there is built.
>  
> 
> You should either revert the kdewin build script to checkout the
> source from a git tag (v0.6.4) and not master as done by other
> packages (I'm sorry, I cannot do that because of missing knowledge).
> 
> At the  mentioned mingw cross build package for kdelib4support I
> need to add  -DHAVE_INET_PTON=1 -DHAVE_INET_NTOP=1 to the cmake
> configure as workaround to the broken inet_xxx function detection
> inside kdeblibs4support source.
> 
> Sorry, but that is not an option - the CI system builds the same set of
> branches that a developer and our release managers build.
> That means 'master'.
> 
That seems not be true for all packages, see
https://invent.kde.org/packaging/craft-blueprints-kde/-/blob/master/kdesupport/qca/qca.py#L17


> The fact that you have to apply additional overrides here indicates to
> me that your changes to 'kdewin' are also incorrect.

It may also be caused by the kdelibs4support package, as I initial
stated - I double checked in kdewin git repo, that it exports the
correct header and library functions.

Ralf
> 
>>
>> -- Forwarded message -
>> From: *CI System* mailto:nore...@kde.org>>
>> Date: Sat, Jul 10, 2021 at 7:21 AM
>> Subject: KDE CI: Administration » Dependency Build Applications
>> stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 52 - Still Failing!
>> To: mailto:bcooks...@kde.org>>
>>
>>
>>  *BUILD FAILURE*
>> Build URL
>> 
>> https://build.kde.org/job/Administration/job/Dependency%20Build%20Applications%20stable-kf5-qt5%20WindowsMSVCQt5.15/52/
>> 
>> <https://build.kde.org/job/Administration/job/Dependency%20Build%20Applications%20stable-kf5-qt5%20WindowsMSVCQt5.15/52/>
>>
>> Project: Dependency Build Applications stable-kf5-qt5
>> WindowsMSVCQt5.15
>> Date of build:   Fri, 09 Jul 2021 18:58:42 +
>> Build duration:  22 min and counting
>>
>>
>> *CONSOLE OUTPUT *
>> [...truncated 51900 lines...]
>> [2021-07-09T19:20:03.112Z]
>> 
>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>> warning C4005: 'Q_OS_WIN': macro redefinition
>> [2021-07-09T19:20:03.112Z] ..\src\kdeui\kicon.cpp: note: see
>> previous definition of 'Q_OS_WIN'
>> [2021-07-09T19:20:03.112Z] [153/533] Building CXX object
>> src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kdialogqueue.cpp.obj
>> [2021-07-09T19:20:03.112Z]
>> 
>> C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
>> warning C4005: 'Q_OS_WIN': macro redefinition
>> [2021-07-09T19:20:03.112Z] ..\src\kdeui\kdialogqueue.cpp: note:
>> see previous definition of 'Q_OS_WIN'
>> [2021-07-09T19:20:03.112Z] [154/533]

Re: Fwd: KDE CI: Administration » Dependency Build Applications stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 52 - Still Failing!

[Ralf Habacker]

Am 09.07.21 um 21:26 schrieb Ben Cooksley:

Hi Ralf,

I'm afraid your previous fixes are insufficient - we are still failing 
with a linking error.


The following is notice that I intend to back all changes out in 
kdewin and restore the repository to 
8503ac6e0e07099c4938cbfe60e3f5e25e2ec368 to restore the correct 
function of the CI system in 24 hours time.


Please don't that not - that fix is required for building 
kdelibs4support with mingw headers version 9 and only export replacement 
functions not provided by the os.


You should either revert the kdewin build script to checkout the source 
from a git tag (v0.6.4) and not master as done by other packages (I'm 
sorry, I cannot do that because of missing knowledge).


At the  mentioned mingw cross build package for kdelib4support I need to 
add  -DHAVE_INET_PTON=1 -DHAVE_INET_NTOP=1 to the cmake configure as 
workaround to the broken inet_xxx function detection inside 
kdeblibs4support source.


Ralf


Regards,
Ben

-- Forwarded message -
From: *CI System* mailto:nore...@kde.org>>
Date: Sat, Jul 10, 2021 at 7:21 AM
Subject: KDE CI: Administration » Dependency Build Applications 
stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 52 - Still Failing!

To: mailto:bcooks...@kde.org>>


*BUILD FAILURE*
Build URL 
https://build.kde.org/job/Administration/job/Dependency%20Build%20Applications%20stable-kf5-qt5%20WindowsMSVCQt5.15/52/ 
 


Project:Dependency Build Applications stable-kf5-qt5 WindowsMSVCQt5.15
Date of build:  Fri, 09 Jul 2021 18:58:42 +
Build duration: 22 min and counting


*CONSOLE OUTPUT *
[...truncated 51900 lines...]
[2021-07-09T19:20:03.112Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.112Z] ..\src\kdeui\kicon.cpp: note: see previous 
definition of 'Q_OS_WIN'
[2021-07-09T19:20:03.112Z] [153/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kdialogqueue.cpp.obj
[2021-07-09T19:20:03.112Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.112Z] ..\src\kdeui\kdialogqueue.cpp: note: see 
previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:03.112Z] [154/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\ktzfiletimezone.cpp.obj
[2021-07-09T19:20:03.112Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.112Z] ..\src\kdecore\ktzfiletimezone.cpp: note: 
see previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:03.112Z] [155/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kapplication.cpp.obj
[2021-07-09T19:20:03.112Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.112Z] ..\src\kdeui\kapplication.cpp: note: see 
previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:03.684Z] [156/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kstringvalidator.cpp.obj
[2021-07-09T19:20:03.684Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.684Z] ..\src\kdeui\kstringvalidator.cpp: note: 
see previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:03.684Z] [157/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kfadewidgeteffect.cpp.obj
[2021-07-09T19:20:03.684Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.684Z] ..\src\kdeui\kfadewidgeteffect.cpp: note: 
see previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:03.945Z] [158/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kbuttongroup.cpp.obj
[2021-07-09T19:20:03.945Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:03.945Z] ..\src\kdeui\kbuttongroup.cpp: note: see 
previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:04.205Z] [159/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\kaction.cpp.obj
[2021-07-09T19:20:04.205Z] 
C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189): 
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-09T19:20:04.205Z] ..\src\kdeui\kaction.cpp: note: see 
previous definition of 'Q_OS_WIN'
[2021-07-09T19:20:04.205Z] [160/533] Building CXX object 
src\CMakeFiles\KF5KDELibs4Support.dir\kdeui\ksessionmanager.cpp.obj
[2021-07-09T19:20:04.205Z] 

Re: KDE CI: Administration » Dependency Build Extragear stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 53 - Still Failing!

[Ralf Habacker]

Am 08.07.21 um 20:17 schrieb Ben Cooksley:
Hi Ben,
On Fri, Jul 9, 2021 at 1:28 AM Harald Sitter > wrote:


comparing the failing build to the last successful one it appears
kdewin no longer installs arpa/inet.h which lead me to this commit

https://invent.kde.org/packaging/kdewin/-/commit/25a254ab3ab0369e235db7418ac9b95226445a38


which seems a bit folly because it makes the installation
conditional on inet_ntop being available with a **different**
header rendering the change not particularly backwards compatible
because of course kdelibs4support is using the old header. One way
forward would certainly be to port kdelibs4support to an ifdef
q_os_win32 that then includes winsock2 instead of the posix
headers. Ralf will probably know best what to do.


Ralf, please note that this needs to be fixed with urgency as it is 
preventing the Windows CI system from being maintainable.


I fixed that issue yesterday, tagged a new release 0.6.6 (see 
https://invent.kde.org/packaging/kdewin/-/tags/v0.6.6) and got 
successful builds with that (see for example 
https://build.opensuse.org/package/live_build_log/windows:mingw:win64/mingw64-kdelibs4support/openSUSE_Leap_15.2/x86_64)


If these changes have not reached KDE CI, it could be because the cached 
kdewin binary package is outdated and needs to be renewed. But I can't 
tell you how to do that.


Regards
Ralf


Regards,
Ben


On Thu, Jul 8, 2021 at 8:46 AM Ben Cooksley mailto:bcooks...@kde.org>> wrote:

Hi all,

Please find below a build log from a Windows dependency build
job which is currently failing in kdelibs4support.

Any ideas what may have caused this?

Thanks,
Ben

-- Forwarded message -
From: *CI System* mailto:nore...@kde.org>>
Date: Thu, Jul 8, 2021 at 8:46 AM
Subject: KDE CI: Administration » Dependency Build Extragear
stable-kf5-qt5 WindowsMSVCQt5.15 - Build # 53 - Still Failing!
To: mailto:bcooks...@kde.org>>


*BUILD FAILURE*
Build URL

https://build.kde.org/job/Administration/job/Dependency%20Build%20Extragear%20stable-kf5-qt5%20WindowsMSVCQt5.15/53/



Project:Dependency Build Extragear stable-kf5-qt5
WindowsMSVCQt5.15
Date of build:  Wed, 07 Jul 2021 18:00:04 +
Build duration: 2 hr 46 min and counting


*CONSOLE OUTPUT *
[...truncated 53095 lines...]
[2021-07-07T20:42:36.730Z] [84/533] Building CXX object
src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\kcurrencycode.cpp.obj

[2021-07-07T20:42:36.730Z]

C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-07T20:42:36.730Z] ..\src\kdecore\kcurrencycode.cpp:
note: see previous definition of 'Q_OS_WIN'
[2021-07-07T20:42:37.710Z] [85/533] Building CXX object
src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\kdebugdbusiface.cpp.obj

[2021-07-07T20:42:37.710Z]

C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-07T20:42:37.710Z] ..\src\kdecore\kdebugdbusiface.cpp:
note: see previous definition of 'Q_OS_WIN'
[2021-07-07T20:42:37.974Z] [86/533] Building CXX object
src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\kdebug.cpp.obj
[2021-07-07T20:42:37.974Z]

C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-07T20:42:37.974Z] ..\src\kdecore\kdebug.cpp: note:
see previous definition of 'Q_OS_WIN'
[2021-07-07T20:42:39.393Z] [87/533] Building CXX object
src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\klibrary.cpp.obj
[2021-07-07T20:42:39.393Z]

C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-07T20:42:39.394Z] ..\src\kdecore\klibrary.cpp: note:
see previous definition of 'Q_OS_WIN'
[2021-07-07T20:42:39.394Z] [88/533] Building CXX object
src\CMakeFiles\KF5KDELibs4Support.dir\kdecore\k4aboutdata.cpp.obj
[2021-07-07T20:42:39.394Z]

C:\Craft\CI-Qt515\windows-msvc2019_64-cl-debug\include\qt5\QtCore/qsystemdetection.h(189):
warning C4005: 'Q_OS_WIN': macro redefinition
[2021-07-07T20:42:39.394Z] ..\src\kdecore\k4aboutdata.cpp:
note: see previous definition of 'Q_OS_WIN'

RE: RE: Broken build for Alkimia

[ralf . habacker]

  Hi Ben,

As quick resolve I can see adding that dependency as one solution


the mac builds failed because of missing
04:27:54  ld: library not found for -lWebKitWidgets

On windows: 


Could not find a package configuration file provided by "Qt5WebKitWidgets"


and the third options is to switch to the experimental QWebEngine


Below you can see the supported and not supported features.
   Fetching pages with javascript is not supported
   Preview of fetched pages is supported
   Multilingual preview of web pages is not supported
   Inspecting of web pages is not supported


Regards
Ralf
-ursprngliche Nachricht-
Von:  ralf.habac...@freenet.de (mailto:ralf.habac...@freenet.de; target="_blank" 
rel="noopener)
Gesendet: 18.09.2020 20:35 Uhr
An:  bcooks...@kde.org (mailto:bcooks...@kde.org; target="_blank" rel="noopener),  ralf.habac...@freenet.de 
(mailto:ralf.habac...@freenet.de; target="_blank" rel="noopener),  r...@habacker.de (mailto:r...@habacker.de; 
target="_blank" rel="noopener),  m...@dawidwrobel.com (mailto:m...@dawidwrobel.com; target="_blank" rel="noopener)
Kopie:  kde-devel@kde.org (mailto:kde-devel@kde.org; target="_blank" 
rel="noopener)
Betreff: RE: Broken build for Alkimia



Hi Ben,
thanks for this pointer.

The changes in master were related to the remove of kdelibs4support, which has been requested recently by Thomas Baumgart. Unfortunally with the removal of kdelibs4support and the fact that alkimia still depends on WebKit because QWebEngine suppport is experimental 
https://invent.kde.org/office/alkimia/-/blob/master/CMakeLists.txt#L83, there is now a dependency to libKF5WebKit required, which seems to not been added to the build system.


As quick resolve I can see adding that dependency as one solution,  another 
other would be that I revert my changes and got back to kdelib4support and the 
third options is to switch to the experimental QWebEngine support, by adding 
-DBUILD_WITH_WEBENGINE=1 to the cmake configure line.

Regards
Ralf

-ursprngliche Nachricht-
Von:  bcooks...@kde.org (mailto:bcooks...@kde.org; target="_blank" 
rel="noopener)
Gesendet: 18.09.2020 18:08 Uhr
An:  ralf.habac...@freenet.de (mailto:ralf.habac...@freenet.de; target="_blank" rel="noopener),  r...@habacker.de 
(mailto:r...@habacker.de; target="_blank" rel="noopener),  m...@dawidwrobel.com (mailto:m...@dawidwrobel.com; 
target="_blank" rel="noopener)
Kopie:  kde-devel@kde.org (mailto:kde-devel@kde.org; target="_blank" 
rel="noopener)
Betreff: Broken build for Alkimia


Hi Ralf and Dawid,

Recently you both made commits/merged in a series of changes to 'Akimia' which 
has resulted in it's CI builds failing now on all platforms.

If you could please correct these failures that would be appreciated - they are 
currently causing Dependency Builds to fail, which means no changes may be made 
to any part of the CI system (otherwise the entire system is broken for 
everyone)

Regards,
Ben


-ursprngliche Nachricht Ende-



-ursprngliche Nachricht Ende-

RE: Broken build for Alkimia

[ralf . habacker]

  Hi Ben, 
thanks for this pointer.


The changes in master were related to the remove of kdelibs4support, which has been requested recently by Thomas Baumgart. Unfortunally with the removal of kdelibs4support and the fact that alkimia still depends on WebKit because QWebEngine suppport is experimental 
https://invent.kde.org/office/alkimia/-/blob/master/CMakeLists.txt#L83, there is now a dependency to libKF5WebKit required, which seems to not been added to the build system.


As quick resolve I can see adding that dependency as one solution,  another 
other would be that I revert my changes and got back to kdelib4support and the 
third options is to switch to the experimental QWebEngine support, by adding 
-DBUILD_WITH_WEBENGINE=1 to the cmake configure line.

Regards
Ralf

-ursprngliche Nachricht-
Von:  bcooks...@kde.org (mailto:bcooks...@kde.org; target="_blank" 
rel="noopener)
Gesendet: 18.09.2020 18:08 Uhr
An:  ralf.habac...@freenet.de (mailto:ralf.habac...@freenet.de; target="_blank" rel="noopener),  r...@habacker.de 
(mailto:r...@habacker.de; target="_blank" rel="noopener),  m...@dawidwrobel.com (mailto:m...@dawidwrobel.com; 
target="_blank" rel="noopener)
Kopie:  kde-devel@kde.org (mailto:kde-devel@kde.org; target="_blank" 
rel="noopener)
Betreff: Broken build for Alkimia


Hi Ralf and Dawid,

Recently you both made commits/merged in a series of changes to 'Akimia' which 
has resulted in it's CI builds failing now on all platforms.

If you could please correct these failures that would be appreciated - they are 
currently causing Dependency Builds to fail, which means no changes may be made 
to any part of the CI system (otherwise the entire system is broken for 
everyone)

Regards,
Ben


-ursprngliche Nachricht Ende-

Re: Gitlab, Git History and Merge Options

[Ralf Habacker]

Am 19.05.20 um 10:10 schrieb Tomaz Canabrava:
By default Gitlab will create merge commits, and this can create a 
really nasty history if  a branch is based on a branch and not in 
master. To overcome this, please - if you own a repository - go to the 
repository settings, Merge Requests, Fast Forward Merge.
It seems that normal invent.kde.org users do not have a menu entry 
"Repository Settings" - is my account missing some rights ?


Ralf

D5173: Fix 'Installation of ksendbugmail.exe conflicts with related KDE4 package'.

[Ralf Habacker]

habacker added a comment.


  Looking into the bug report gaves the answer:
  
file /usr/i686-w64-mingw32/sys-root/mingw/bin/ksendbugmail.exe from install 
of mingw32-libKF5XmlGui5-5.26.0-3.1.noarch conflicts with file from package 
mingw32-kdelibs4-4.14.60-11.4.noarch
  
  For the recored: this could happens on obs, where stable releases for 
umbrello and kmymoney on Windows are build.

REPOSITORY
  R263 KXmlGui

REVISION DETAIL
  https://phabricator.kde.org/D5173

To: habacker, dfaure, ltoscano, bcooksley
Cc: kde-frameworks-devel, bcooksley, dfaure, aacid, LeGast00n, cblack, 
michaelh, ngraham, bruns

D14598: Keep checking return value from calling browse_stat_path() in SMBSlave::stat() in sync with SMBSlave::listDir()

[Ralf Habacker]

habacker closed this revision.
habacker added a comment.


  This review request was superseeded by D16272 
.

REPOSITORY
  R320 KIO Extras

REVISION DETAIL
  https://phabricator.kde.org/D14598

To: habacker, dfaure
Cc: kde-frameworks-devel, kfm-devel, azyx, nikolaik, pberestov, iasensio, 
fprice, LeGast00n, cblack, fbampaloukas, alexde, GB_2, Codezela, feverfew, 
meven, michaelh, spoorun, navarromorales, firef, ngraham, andrebarros, bruns, 
emmanuelp, rdieter, mikesomov

D21660: remove dbus for windows build and change audio dep logic

[Ralf Habacker]

habacker added a comment.


  In D21660#565205 , @cullmann wrote:
  
  > That is bad, then this must be fixed.
  >  That should happen on Windows, too.
  
  
  BTW: Does starting slaves on Linux kills started slave after slave timeout 
occurs, which is performed by klauncher by default ? 
  In case you are running a KDE app  using kioslave on a removal disc, after 
closing the main app, you will have a problem ejecting the removal disc because 
of still running kioslaves.

REPOSITORY
  R289 KNotifications

REVISION DETAIL
  https://phabricator.kde.org/D21660

To: brute4s99, broulik, nicolasfella
Cc: cullmann, habacker, aspotashev, bcooksley, apol, nicolasfella, 
kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns

D21660: remove dbus for windows build and change audio dep logic

[Ralf Habacker]

habacker added a comment.


  In D21660#565113 , @cullmann wrote:
  
  > No, not really, I fixed that long ago, we auto-detect this and use the 
KDE_FORK_SLAVES code paths for that
  >
  > static bool forkSlaves()
  
  
  but this function is only called on Linux 
(https://github.com/KDE/kio/blob/a8de3cb2032aba3dc86b1c7a7e745ca000f21bf4/src/core/slave.cpp#L458),
 so my statement is still true.
  
#ifdef Q_OS_UNIX
if (forkSlaves() == 1) {
  ...
  QProcess::startDetached(kioslave, args);

return slave;
}
#endif
  
  Later on there is the code, called on Windows, which requires klauncher
  
QString errorStr;
  QDBusReply reply = klauncher()->requestSlave(protocol, url.host(), 
slaveAddress.toString(), errorStr);
  if (!reply.isValid()) {
  error_text = i18n("Cannot talk to klauncher: %1", 
klauncher()->lastError().message());
  error = KIO::ERR_CANNOT_LAUNCH_PROCESS;
  delete slave;
  return 0;

REPOSITORY
  R289 KNotifications

REVISION DETAIL
  https://phabricator.kde.org/D21660

To: brute4s99, broulik, nicolasfella
Cc: cullmann, habacker, aspotashev, bcooksley, apol, nicolasfella, 
kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns

D21660: remove dbus for windows build and change audio dep logic

[Ralf Habacker]

habacker added a comment.


  In D21660#565110 , @cullmann wrote:
  
  > kio doesn't require dbus for the normal slave stuff.
  >  It can be that it requires it for the kcookie or auth stuff.
  
  
  klauncher is used to launch kio slaves. In the main() function of klauncher 
there is a check if a DBus session bus is running 
(https://lxr.kde.org/source/frameworks/kinit/src/klauncher/klauncher_main.cpp#0187)
 . If not, klauncher is terminated. This looks like a hard requirement for DBus 
if kio slaves are used.

REPOSITORY
  R289 KNotifications

REVISION DETAIL
  https://phabricator.kde.org/D21660

To: brute4s99, broulik, nicolasfella
Cc: cullmann, habacker, aspotashev, bcooksley, apol, nicolasfella, 
kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns

D21660: remove dbus for windows build and change audio dep logic

[Ralf Habacker]

habacker added a comment.


  In D21660#563560 , @bcooksley 
wrote:
  
  > Considering that D-Bus doesn't really belong on Windows,
  
  
  KDE uses dbus for many internal services. On a recent linux system I see the 
following KDE services in dbus
  
  org.kde.ActivityManager
   org.kde.JobViewServer
   org.kde.kappmenu
   org.kde.kcookiejar5
   org.kde.kdeconnectd
   org.kde.kded5
   org.kde.keyboard
   org.kde.kglobalaccel
   org.kde.kiod5
   org.kde.kioexecd
   org.kde.klauncher5
   org.kde.klipper
   org.kde.knotify
   org.kde.kpasswdserver
   org.kde.krunner
   org.kde.kssld5
   org.kde.kuiserver
   org.kde.kwalletd5
   org.kde.Solid.PowerManagement
   org.kde.Solid.PowerManagement.PolicyAgent
   org.kde.StatusNotifierWatcher
  
  Some of them are Linux only, but some are required on Windows (for example 
kio)
  
  > and has been known to cause security software (such as anti-malware 
packages) to generate false positives in the past,
  
  This has not been reported to the official dbus issue tracker 
(https://gitlab.freedesktop.org/groups/dbus/-/issues?scope=all=%E2%9C%93=opened=virus)
 . Can you point me to related reports ?
  
  > For standalone applications there isn't really a reason to have it around.
  
  From what I know does kio depends on dbus, which mean you are saying that 
standalone applications could not use kio and my be others (see above)

REPOSITORY
  R289 KNotifications

REVISION DETAIL
  https://phabricator.kde.org/D21660

To: brute4s99, broulik, nicolasfella
Cc: cullmann, habacker, aspotashev, bcooksley, apol, nicolasfella, 
kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns

D21660: remove dbus for windows build and change audio dep logic

[Ralf Habacker]

habacker added a comment.


  In D21660#563425 , @bcooksley 
wrote:
  
  > If tests launch D-Bus Daemon, then they will cause the CI system to jam as 
CTest will wait indefinitely for dbus-daemon to exit.
  
  
  There is a fix for this issue, see https://phabricator.kde.org/D25350

REPOSITORY
  R289 KNotifications

REVISION DETAIL
  https://phabricator.kde.org/D21660

To: brute4s99, broulik, nicolasfella
Cc: habacker, aspotashev, bcooksley, apol, nicolasfella, kde-frameworks-devel, 
LeGast00n, GB_2, michaelh, ngraham, bruns

D21660: remove dbus for windows build and change audio dep logic

[Ralf Habacker]

habacker added a comment.


  In D21660#481421 , @bcooksley 
wrote:
  
  > With regards to Windows, please note that any unit test which depends on 
calls that involve D-Bus on the CI system will likely lead to that test hanging 
because dbus-daemon is not launched by the CI system.
  
  
  There is no need to start dbus-daemon by CI. dbus-daemon is started by the 
dbus client library by default, if not running for the given install path.

REPOSITORY
  R289 KNotifications

REVISION DETAIL
  https://phabricator.kde.org/D21660

To: brute4s99, broulik, nicolasfella
Cc: habacker, aspotashev, bcooksley, apol, nicolasfella, kde-frameworks-devel, 
LeGast00n, GB_2, michaelh, ngraham, bruns

D12283: Allow to load dictionaries form a rcc

[Ralf Habacker]

habacker added a comment.
Herald edited subscribers, added: kde-frameworks-devel; removed: Frameworks.


  In D12283#249520 , @vonreth wrote:
  
  > While the disk use of the installation goes down the ram usage explodes as 
QResource doesn't use mmap 
https://github.com/qt/qtbase/blob/a37dd93defd91b79fb6730d0ff0515a66a0d3972/src/corelib/io/qresource.cpp#L1037
  
  
  Would adding a mmap implementation on Windows like 
https://github.com/witwall/mman-win32 be an option  ?

REPOSITORY
  R246 Sonnet

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D12283

To: vonreth, mlaurent
Cc: kde-frameworks-devel, habacker, LeGast00n, GB_2, michaelh, ngraham, bruns, 
#frameworks

D19504: Extend KHtmlView::print() to use a predefined QPrinter instance

[Ralf Habacker]

This revision was automatically updated to reflect the committed changes.
Closed by commit R313:2d8cba050729: Extend KHtmlView::print() to use a 
predefined QPrinter instance (authored by habacker).

CHANGED PRIOR TO COMMIT
  https://phabricator.kde.org/D19504?vs=67894=68479#toc

REPOSITORY
  R313 KHtml

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D19504?vs=67894=68479

REVISION DETAIL
  https://phabricator.kde.org/D19504

AFFECTED FILES
  src/khtmlview.cpp
  src/khtmlview.h

To: habacker, dfaure, vkrause
Cc: aacid, tbaumgart, kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, 
bruns

D19504: Extend KHtmlView::print() to use a predefined QPrinter instance

[Ralf Habacker]

habacker updated this revision to Diff 67894.
habacker added a comment.


  - Another attempt after 9 minor releases have elapsed.

REPOSITORY
  R313 KHtml

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D19504?vs=53327=67894

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D19504

AFFECTED FILES
  src/khtmlview.cpp
  src/khtmlview.h

To: habacker
Cc: aacid, tbaumgart, kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, 
bruns

D19504: Extend KHtmlView::print() to use a predefined QPrinter instance

[Ralf Habacker]

habacker added a comment.


  ping ?

REPOSITORY
  R313 KHtml

REVISION DETAIL
  https://phabricator.kde.org/D19504

To: habacker
Cc: aacid, tbaumgart, kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, 
bruns

Re: CI system maintainability

[Ralf Habacker]

Am 29.03.19 um 21:01 schrieb Ben Cooksley:
> With the shift to Gitlab we should be able to provide this hopefully.
>
> We're still figuring out how to be able to provide CI in an easy to
> maintain manner (in terms of controlling platforms builds are done
> for, which branches, etc).

In case, an example is wanted:
https://gitlab.freedesktop.org/dbus/dbus/blob/master/.gitlab-ci.yml
specifies a working configuration for building several builds variants
for linux and cross compiled Windows.

Ralf

Re: CI system maintainability

[Ralf Habacker]

Am 29.03.19 um 21:01 schrieb Ben Cooksley:
> With the shift to Gitlab we should be able to provide this hopefully.
>
> We're still figuring out how to be able to provide CI in an easy to
> maintain manner (in terms of controlling platforms builds are done
> for, which branches, etc).

In case, an example is wanted:
https://gitlab.freedesktop.org/dbus/dbus/blob/master/.gitlab-ci.yml
specifies a working configuration for building several builds variants
for linux and cross compiled Windows.

Ralf

Re: CI system maintainability

[Ralf Habacker]

Am 29.03.19 um 21:01 schrieb Ben Cooksley:
> With the shift to Gitlab we should be able to provide this hopefully.
>
> We're still figuring out how to be able to provide CI in an easy to
> maintain manner (in terms of controlling platforms builds are done
> for, which branches, etc).

In case, an example is wanted:
https://gitlab.freedesktop.org/dbus/dbus/blob/master/.gitlab-ci.yml
specifies a working configuration for building several builds variants
for linux and cross compiled Windows.

Ralf

D19504: Extend KHtmlView::print() to use a predefined QPrinter instance

[Ralf Habacker]

habacker updated this revision to Diff 53327.
habacker marked an inline comment as done.
habacker added a comment.


  - added @since tag
  
  The patch was added to build the stable version of KMyMoney
  for Windows to fix an issue not been able to have common
  printer settings. Since KHTML is the successor of kdelilbs,
  it has been ported to KF5.

REPOSITORY
  R313 KHtml

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D19504?vs=53104=53327

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D19504

AFFECTED FILES
  src/khtmlview.cpp
  src/khtmlview.h

To: habacker
Cc: aacid, tbaumgart, kde-frameworks-devel, michaelh, ngraham, bruns

D19504: Extend KHtmlView::print() to use a predefined QPrinter instance

[Ralf Habacker]

habacker marked an inline comment as done.
habacker added inline comments.

INLINE COMMENTS

> tbaumgart wrote in khtmlview.cpp:3134
> Why don't you pass a reference to the QPrinter object here?

Because api  dealing with QPrinter always use a pointer.

void QWebView::print(QPrinter *)
void QPrintPreviewDialog::paintRequested(QPrinter *printer)
QPrintPreviewDialog(QPrinter *printer, QWidget *parent = nullptr, 
Qt::WindowFlags flags = Qt::WindowFlags()
QPrintDialog(QPrinter * printer, QWidget * parent = 0)
QPageSetupDialog(QPrinter *printer, QWidget *parent)
QPrintPreviewWidget(QPrinter *printer, QWidget *parent, Qt::WindowFlags flags)

REPOSITORY
  R313 KHtml

REVISION DETAIL
  https://phabricator.kde.org/D19504

To: habacker
Cc: tbaumgart, kde-frameworks-devel, michaelh, ngraham, bruns

D19504: Extend KHtmlView::print() to use a predefined QPrinter instance

[Ralf Habacker]

habacker created this revision.
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  This is required to share printer settings inside an application.
  
  BUG:405011

TEST PLAN
  compile on linux

REPOSITORY
  R313 KHtml

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D19504

AFFECTED FILES
  src/khtmlview.cpp
  src/khtmlview.h

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D17650: Install kioslave.exe as kioslave5.exe under Windows

[Ralf Habacker]

habacker added a comment.


  In D17650#380309 , @habacker wrote:
  
  > See this https://bugs.kde.org/show_bug.cgi?id=380139 as example. The 
different Qt installation layout depending on the platform is a nightmare for 
cross-platform support and ties up resources that are no longer available for 
more important things.
  
  
  Here you can an actual example - https://phabricator.kde.org/T10207

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D17650: Install kioslave.exe as kioslave5.exe under Windows

[Ralf Habacker]

habacker updated this revision to Diff 47944.
habacker retitled this revision from "Install kioslave as kioslave5 on Windows" 
to "Install kioslave.exe as kioslave5.exe under Windows".
habacker edited the summary of this revision.
habacker removed subscribers: bcooksley, heikobecker, ngraham, lbeltrame.
habacker added a comment.


  - fix clients of kioslave

REPOSITORY
  R241 KIO

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D17650?vs=47757=47944

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D17650

AFFECTED FILES
  src/core/slave.cpp
  src/kioslave/CMakeLists.txt

To: habacker, #frameworks, lbeltrame, vonreth
Cc: kde-frameworks-devel, michaelh, ngraham, bruns, bcooksley, heikobecker

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  In D17650#380277 , @habacker wrote:
  
  > and try to fix the related issue or check if some else has fixed it.
  
  
  See this https://bugs.kde.org/show_bug.cgi?id=380139 as example. The 
different Qt installation layout depending on the platform is a nightmare for 
cross-platform support and ties up resources that are no longer available for 
more important things. Making cross compiled packaged on obs for example is 
mostly a simple copy and replace of specific variables (for example that shared 
files are located in bin dir, not in lib). Converting to the complete different 
Qt 5 path layout results into much more work without any benefit.

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: bcooksley, heikobecker, ngraham, lbeltrame, kde-frameworks-devel, michaelh, 
bruns

D17719: Fix executable name for kioslave on Windows

[Ralf Habacker]

habacker created this revision.
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  FIXED-IN:5.54.0
  BUG:377687
  also see https://phabricator.kde.org/D17650

REPOSITORY
  R303 KInit

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D17719

AFFECTED FILES
  src/klauncher/klauncher.cpp

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  In D17650#380272 , @vonreth wrote:
  
  > In kde4 we had 104 unmaintained untested applications.
  >  The 6 now are much better supported, in general we only add applications 
to binary factory if a project asks for it and maintains it for some degree.
  
  
  
  
  In D17650#380266 , @heikobecker 
wrote:
  
  > > no, it is only used in kio internally by klauncher to start io slaves.
  >
  > at least kinit disagrees:
  >
  >   src/klauncher/klauncher.cpp:1021
arg_list.prepend(QLatin1String("kioslave"));
  >
  
  
  Thanks for this pointer, which is one of the major differences to KDE4 and 
makes it sometime hard to follow. In KDE4 kinit  is also located in the same 
repo.

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: bcooksley, heikobecker, ngraham, lbeltrame, kde-frameworks-devel, michaelh, 
bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  > Why not joining the effort of the people that are working to get those 
applications on Windows,
  
  You did not get the point. If there is anything left over besides the time I 
have to maintain the KDE applications and libraries, I go through this list 
(https://bugs.kde.org/showdependencytree.cgi?id=373932_resolved=1) and try 
to fix the related issue.
  The list has already been reduced from 66 to 28.

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: bcooksley, heikobecker, ngraham, lbeltrame, kde-frameworks-devel, michaelh, 
bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  Yes there are more 'nightly' releases, which is a good thing :-),
  
  but there are only about  6 'release' versions
  
  https://binary-factory.kde.org/view/Windows%2064-bit/ : 5 
  https://binary-factory.kde.org/view/Windows%2032-bit/ : 6
  
  and KMyMoney 'release' version is marked as 'Preview Release' (see 
https://kmymoney.org/)

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: bcooksley, heikobecker, ngraham, lbeltrame, kde-frameworks-devel, michaelh, 
bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  > I'd argue that the problem is with your distribution.
  
  It is designed in ECM to install mentioned helpers on windows (which are 
cross compiled packages on obs) in main executable install dir.
  
$ rpm -q -f /usr/i686-w64-mingw32/sys-root/mingw/bin/kioslave.exe 
mingw32-kdelibs4-4.14.60-30.27.noarch
$ rpm -q -f /usr/i686-w64-mingw32/sys-root/mingw/bin/kioslave.exe 
mingw32-kio-5.47.0-11.28.noarch
  
  > Everybody still shipping qt4 is on its own anyway,
  
  Unfortunally yes, for example 
https://build.opensuse.org/package/show/windows%3Amingw%3Awin32/mingw32-kdelibs4,
 but this is unrelated, we are about talking about a patch for KF5, not qt4.
  
  > seeing that it's not possible to upstream patches (and there are quite a 
few needed to keep it building) as the git repo is locked down.
  
  It needs someone to take over maintenance of kdelibs.
  
  > And I'm pretty sure this patch breaks at least one other framework.
  
  no, it is only used in kio internally by klauncher to start io slaves.
  
  > Additionally there might be future maintenance costs,
  
  no, it simply follows the name style other kio executables installed in 
normal 'bin' dir already have e.g kcookiejar5, ktelnetservices5, ktrash5.
  
  > so I see no reason to flog a dead horse and penalize everybody else for it.
  
  In recent years, of the 104 KDE4 applications for Windows 
(https://community.kde.org/Windows/Releases/4.10.2), only 7 are available for 
KF5 (https://community.kde.org/Windows, minus the stable versions of Umbrello 
and KMymoney that are still on KDE4).  KDE4 will probably stay as a base for a 
while.

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: heikobecker, ngraham, lbeltrame, kde-frameworks-devel, michaelh, bruns

D17649: Let docbookl10nhelper executable name follow conventions of checkXML on Windows

[Ralf Habacker]

habacker added a comment.


  On opensuse there are the cross compile helper packages
  
  mingw32-cross-kde4-tools
  mingw32-cross-kf5-tools
  
  both containing docbookl10nhelper and nothing prevents them from been 
installed in parallel

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D17649

To: habacker
Cc: ltoscano, kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, 
skadinna

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  I think this is unrelated - this request is to fix an issue with an available 
package on a distribution, so can anyone accept this ?

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: ngraham, lbeltrame, kde-frameworks-devel, michaelh, bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  The conflict is still presence on opensuse and relates to the packages
  
  mingw32-cross-kde4-tools
  mingw32-cross-kf5-tools

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: lbeltrame, kde-frameworks-devel, michaelh, ngraham, bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  On opensuse there are the cross compile helper packages
  
  mingw32-cross-kde4-tools 
  mingw32-cross-kf5-tools
  
  both containing docbookl10nhelper and nothing prevents them from been 
installed in parallel

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: lbeltrame, kde-frameworks-devel, michaelh, ngraham, bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  On a recent opensuse Leap 42.3 or 15.x system there is
  
  > /usr/lib64/kde4/libexec/kioslave
  >  /usr/lib64/libexec/kf5/kioslave
  
  because the distribution provides kdelibs4 and kio packages. it only does not 
conflict because the executables are located in different directories because 
they are installed in 'libexecdir'. On Windows libexecdir points to the regular 
bin dir, which results into a conflict. The conflict need to be fixed by 
choosing a different name, as it is done also with many helper tools like 
meinproc4, meinproc5
  
  > $ ls /usr/bin/*5 | wc -l
  > 73
  >  $ls /usr/bin/*4 | wc -l
  > 46

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame, vonreth
Cc: lbeltrame, kde-frameworks-devel, michaelh, ngraham, bruns

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker added a comment.


  In D17650#378747 , @lbeltrame 
wrote:
  
  > At this point, I'd rather give my -1 to this. The 4.x kdelibs stack is 
*long* unmaintained, as well as Qt. What are real, **compelling** reasons to do 
this?
  >
  > EDIT: I know it's for Windows, but with the recent work with the 
binary-factory and what not, there are even less reasons to do so there.
  
  
  umbrello from binary factory is far from been production ready. Windows 
releases are made from the KDE4 branch. See for example 
https://phabricator.kde.org/T7659

REPOSITORY
  R241 KIO

REVISION DETAIL
  https://phabricator.kde.org/D17650

To: habacker, #frameworks, lbeltrame
Cc: lbeltrame, kde-frameworks-devel, michaelh, ngraham, bruns

D17649: Let docbookl10nhelper executable name follow conventions of checkXML on Windows

[Ralf Habacker]

habacker added a comment.


  In D17649#378737 , @ltoscano wrote:
  
  > The file is not meant to be installed (unless INSTALL_INTERNAL_TOOLS is 
exported), and in general it should not exist in the same build root of its 
kdelibs 4.x counterpart.
  
  
  This tools is required for cross compiling kdoctools 
https://build.opensuse.org/package/show/home:rhabacker:branches:windows:mingw:win64:KF547/mingw64-kdoctools
 and therefore installed. obs has KDE4 and KF5 packages side by side

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D17649

To: habacker
Cc: ltoscano, kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, 
skadinna

D17650: Install kioslave as kioslave5 on Windows

[Ralf Habacker]

habacker created this revision.
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  FIXED-IN:5.54.0
  BUG:377687

REPOSITORY
  R241 KIO

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D17650

AFFECTED FILES
  src/kioslave/CMakeLists.txt

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D17649: Let docbookl10nhelper executable name follow conventions of checkXML on Windows

[Ralf Habacker]

habacker created this revision.
Herald added projects: Frameworks, Documentation.
Herald added subscribers: kde-doc-english, kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  FIXED-IN:5.54
  BUG:373928

REPOSITORY
  R238 KDocTools

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D17649

AFFECTED FILES
  src/CMakeLists.txt

To: habacker
Cc: kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, skadinna

D16963: Fix running kdoctools-install autotest without kdoctools installed

[Ralf Habacker]

This revision was automatically updated to reflect the committed changes.
Closed by commit R238:0a714c220662: Fix running kdoctools-install autotest 
without kdoctools installed (authored by habacker).

REPOSITORY
  R238 KDocTools

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D16963?vs=45690=45710

REVISION DETAIL
  https://phabricator.kde.org/D16963

AFFECTED FILES
  KF5DocToolsMacros.cmake
  autotests/CMakeLists.txt

To: habacker, aacid
Cc: asturmlechner, kde-frameworks-devel, kde-doc-english, michaelh, ngraham, 
bruns, skadinna

D16963: Fix running kdoctools-install autotest without kdoctools installed

[Ralf Habacker]

habacker marked an inline comment as done.

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D16963

To: habacker, aacid
Cc: kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, skadinna

D16963: Fix running kdoctools-install autotest without kdoctools installed

[Ralf Habacker]

habacker updated this revision to Diff 45690.
habacker added a comment.


  - fix parallel builds

REPOSITORY
  R238 KDocTools

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D16963?vs=45685=45690

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D16963

AFFECTED FILES
  KF5DocToolsMacros.cmake
  autotests/CMakeLists.txt

To: habacker, aacid
Cc: kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, skadinna

D16963: Fix running kdoctools-install autotest without kdoctools installed

[Ralf Habacker]

habacker added a reviewer: aacid.

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D16963

To: habacker, aacid
Cc: kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, skadinna

D16963: Fix running kdoctools-install autotest without kdoctools installed

[Ralf Habacker]

habacker created this revision.
Herald added projects: Frameworks, Documentation.
Herald added subscribers: kde-doc-english, kde-frameworks-devel.
habacker requested review of this revision.

REPOSITORY
  R238 KDocTools

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D16963

AFFECTED FILES
  KF5DocToolsMacros.cmake
  autotests/CMakeLists.txt

To: habacker
Cc: kde-frameworks-devel, kde-doc-english, michaelh, ngraham, bruns, skadinna

D14785: Fix DialogNativeTest in multi screen configuration

[Ralf Habacker]

habacker added a comment.


  ping

REPOSITORY
  R242 Plasma Framework (Library)

REVISION DETAIL
  https://phabricator.kde.org/D14785

To: habacker, #plasma
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14787: Fix 'ThemeTest::testCompositingChange() fails if windows manager is running'

[Ralf Habacker]

habacker added a comment.


  ping

REPOSITORY
  R242 Plasma Framework (Library)

REVISION DETAIL
  https://phabricator.kde.org/D14787

To: habacker, #plasma
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D15481: Fix 'Impossible to use knsrc file for uploads from non standard location'

[Ralf Habacker]

This revision was automatically updated to reflect the committed changes.
Closed by commit R304:b6a52eef1ab6: Fix Impossible to use knsrc file for 
uploads from non standard location (authored by habacker).

REPOSITORY
  R304 KNewStuff

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D15481?vs=41550=41554

REVISION DETAIL
  https://phabricator.kde.org/D15481

AFFECTED FILES
  src/uploaddialog.cpp

To: habacker, whiting
Cc: whiting, kde-frameworks-devel, michaelh, ngraham, bruns

D15481: Fix 'Impossible to use knsrc file for uploads from non standard location'

[Ralf Habacker]

habacker created this revision.
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  FIXED-IN:5.51
  BUG:397958

TEST PLAN
  build and did run test case on linux

REPOSITORY
  R304 KNewStuff

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D15481

AFFECTED FILES
  src/uploaddialog.cpp

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14940: kformattest: Use plural suffix (s) consequently

[Ralf Habacker]

habacker added a comment.


  
https://forum.wordreference.com/threads/decimals-plural-singular.2054431/#post-10279600
 listed many public organisations that are using signular for values between 1 
and -1.

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14940

To: habacker, aacid
Cc: ngraham, kde-frameworks-devel, michaelh, bruns

D14940: kformattest: Use plural suffix (s) consequently

[Ralf Habacker]

habacker added a comment.


  I see - Qt lacks support for choosing the correct translation string for 
floating point values like this:
  
qreal value = 1.0;
tr("%n second(s)", nullptr, value).
  
  which should return **1.0 second**
  
qreal value = 1.34;
tr("%n second(s)", nullptr, value).
  
  which should return **1.34 seconds**
  
qreal value = 0.0;
tr("%n second(s)", nullptr, value).
  
  which should return **0.0 seconds**
  
  and for strings:
  
int decimalPlaces = 2;
qreal MSecsInSecond = 1000;
qreal msecs = 1000.0
return tr("%n second(s)").arg(m_locale.toString(msecs / (MSecsInSecond * 
1.0), 'f', decimalPlaces));
  
  which should return **1.00 second**
  
int decimalPlaces = 2;
qreal MSecsInSecond = 1000;
qreal msecs = 1100.0
return tr("%n second(s)").arg(m_locale.toString(msecs / (MSecsInSecond * 
1.0), 'f', decimalPlaces));
  
  which should return **1.10 seconds**

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14940

To: habacker, aacid
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14940: kformattest: Use plural suffix (s) consequently

[Ralf Habacker]

habacker added a comment.


  According to the section'Handling of plurals' on 
http://doc.qt.io/qt-5/i18n-source-translation.html, for plurals an '(s)' is 
used instead of the 's' as suffx. You are refering to a different  spec ?

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14940

To: habacker, aacid
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker abandoned this revision.
habacker added a comment.


  superseeded by D14967  (tried to updated 
this review with arc diff, but forgot to add the annoying --verbatim argument 
to let phabricator know that I have a changed commit message :-/)

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks
Cc: dfaure, aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14967: Disable loading of translations for kformattest, which was not designed with translations

[Ralf Habacker]

habacker created this revision.
habacker added reviewers: dfaure, aacid.
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  Required reviews: D14940  D14966 

  
  Review Request: https://phabricator.kde.org/D14779

TEST PLAN
  compiled and checked with kformattest on linux and windows using cross 
compiled build

REPOSITORY
  R244 KCoreAddons

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14967

AFFECTED FILES
  autotests/kformattest.cpp
  src/lib/CMakeLists.txt

To: habacker, dfaure, aacid
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14966: Added support for disabling the loading of translations

[Ralf Habacker]

habacker created this revision.
habacker added reviewers: aacid, dfaure.
Herald added projects: Frameworks, Build System.
Herald added subscribers: kde-buildsystem, kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  With this commit ecm_create_qm_loader() creates a specific header file
  containing related helper functions if NAMESPACE argument is provided.

TEST PLAN
  compiled and tested with patched kcoreaddons from git repo

REPOSITORY
  R240 Extra CMake Modules

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14966

AFFECTED FILES
  modules/ECMPoQmTools.cmake
  modules/ECMQmLoader.cpp.in
  modules/ECMQmLoader.h.in

To: habacker, aacid, dfaure
Cc: kde-frameworks-devel, kde-buildsystem, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  In D14779#311871 , @habacker wrote:
  
  > Thanks for this pointer - with this setting the german translation is not 
fetched anymore.
  
  
  A further look into the windows implementation of the qm loader part 
(https://cgit.kde.org/extra-cmake-modules.git/tree/modules/ECMQmLoader.cpp.in#n65)
 shows that the language is fetched from `QLocale::system() `
  
  QLocale::system() is initiated on unix os from LC_ALL and friends 
(http://code.qt.io/cgit/qt/qtbase.git/tree/src/corelib/tools/qlocale_unix.cpp#n75),
 which could be changed with the qputenv call.
  
  But on Windows the system locale is fetched from the current user locale 
(http://code.qt.io/cgit/qt/qtbase.git/tree/src/corelib/tools/qlocale_win.cpp#n172),
 which could not be changed temporary by the test app.
  
  To have a cross platform solution I think it is really required to disable 
loading translations in this test.  If QLocale::C could not be used for that 
(Albert mentioned that)  an auxiliary function provided by the qmloader is 
required for that.
  I'm going to file a related review request.

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks
Cc: dfaure, aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  Thanks for this pointer - with this setting the german translation is not 
fetched anymore.
  
  But there is still the question open how to deal with plurals with and 
without installed kcoreaddons language package.
  
  I guess running tests without kcoreaddons-lang installed will be performed at 
KDE CI and running tests with kcoreaddons language package mainly happens on 
developers desktop systems, where KF5 is installed and a developer is working 
on kcoreaddons.
  
  From what I read on Qt doc ui string with plurals always need to use a 
translation string with "%n millisecond(s)"  (cleaned with  D14940 
), but in the test code there are two cases
  
QCOMPARE(format.formatDecimalDuration(1 * MSecsInDay + 10 * MSecsInHour, 
3), QStringLiteral("1.417 days"));
  
  This one requires to have a translation installed, which translates "day(s)" 
to "days"
  
  and this one
  
QCOMPARE(format.formatDecimalDuration(10), QStringLiteral("10 
millisecond(s)"));
  
  This fails with a translation installed because an installed english 
translation translates this to "10 milliseconds"
  
  How to proceed ?

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks
Cc: dfaure, aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14940: kformattest: Use plural suffix (s) consequently

[Ralf Habacker]

habacker updated this revision to Diff 40038.
habacker added a comment.


  - type fix in KFormatTest::formatDecimalDuration
  - missing plural fix in KFormatPrivate::formatDecimalDuration

REPOSITORY
  R244 KCoreAddons

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D14940?vs=40027=40038

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14940

AFFECTED FILES
  autotests/kformattest.cpp
  src/lib/util/kformatprivate.cpp

To: habacker, aacid
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  In D14779#311511 , @aacid wrote:
  
  > Can you please replace your setLocale call in initTestCase for
  >
  >   void initLocale()
  >   {
  > setenv("LC_ALL", "en_US.utf-8", 1);
  
  
  Just to inform you: this will not work with msvc 
https://stackoverflow.com/questions/17258029/c-setenv-undefined-identifier-in-visual-studio
 and adds an additional 'kdewin' dependency to windows builds
  
  >   }
  >   Q_CONSTRUCTOR_FUNCTION(initLocale)
  >
  > 
  > This should make it work (or at least it does here)
  
  Just recognized  general issues in kcoreaddons translation, which need to be 
fixed before - please accept D14940 .
  
  Then I will rebase this patch and check your suggestion.

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks
Cc: aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14940: kformattest: Use plural suffix (s) consequently

[Ralf Habacker]

habacker created this revision.
habacker added a reviewer: aacid.
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  Mixing plural '(s)' and translated 's' suffix could not be correct.

REPOSITORY
  R244 KCoreAddons

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14940

AFFECTED FILES
  autotests/kformattest.cpp

To: habacker, aacid
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  In D14779#310116 , @aacid wrote:
  
  > Did you have that problem when you made ecm not create a translator?
  
  
  no,  therefore I created D14778 
  
  > 
  > 
  >> What is the drawback to not load any translation in the ECM qm file loader 
with using QLocale::C as default locale ?
  > 
  > I've seen some systems that default to C locale,
  
  Can you give a pointer to that system ?
  
  > and I would prefer them to still have the english translation working.
  
  but then there need to be a different way to disable translation for this 
test because this
  
  >   QCOMPARE(format.formatSpelloutDuration(361), QStringLiteral("1 
hour(s)"));
  
  and similar lines are not designed to use any translation.

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks
Cc: aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  With the drawback that the qm file need to be generated on each build and the 
po file needs to be updated on any translations change in the test  app
  
  BTW: I tried your approach by
  
  1. downloaded po file from 
https://websvn.kde.org/trunk/l10n-kf5/en/messages/frameworks/kcoreaddons5_qt.po?revision=1522662=markup
  2. did run lrelease on that file and placed it into kcoreaddons source dir as 
kcoreaddons5_qt.qm
  3. added to top level CMakelists.txt
  
  `add_definitions(-DCMAKE_SOURCE_DIR="${CMAKE_SOURCE_DIR}")`
  
  3. add to autotests/kformattest.cpp
  
void KFormatTest::initTestCase()
{
QLocale::setDefault(QLocale::C);
QTranslator *l = new QTranslator();
l->load(QStringLiteral(CMAKE_SOURCE_DIR "/kcoreaddons5_qt.qm"));
QCoreApplication::installTranslator(l);
}
  
  without kcoreaddons-lang package installed the test case returns on a system 
with installed de_DE
  
strace -e trace=file bin/kformattest  2>&1 | grep \.qm
...
access("/home/xxx/src/kf5/kcoreaddons/kcoreaddons5_qt.qm", R_OK) = 0
stat("/home/xxx/src/kf5/kcoreaddons/kcoreaddons5_qt.qm", 
{st_mode=S_IFREG|0644, st_size=729, ...}) = 0
open("/home/xxx/src/kf5/kcoreaddons/kcoreaddons5_qt.qm", 
O_RDONLY|O_CLOEXEC) = 5
  
  
  
~/src/kf5/kcoreaddons-build> bin/kformattest 
* Start testing of KFormatTest *
Config: Using QtTest library 5.11.1, Qt 5.11.1 (x86_64-little_endian-lp64 
shared (dynamic) release build; by GCC 4.8.5)
PASS   : KFormatTest::initTestCase()
PASS   : KFormatTest::formatByteSize()
PASS   : KFormatTest::formatDuration()
FAIL!  : KFormatTest::formatDecimalDuration() Compared values are not the 
same
   Actual   (format.formatDecimalDuration(10))   : "10 milliseconds"
   Expected (QStringLiteral("10 millisecond(s)")): "10 millisecond(s)"
   Loc: [/home/xxx/src/kf5/kcoreaddons/autotests/kformattest.cpp(302)]
FAIL!  : KFormatTest::formatSpelloutDuration() Compared values are not the 
same
   Actual   (format.formatSpelloutDuration(1000)): "1 second"
   Expected (QStringLiteral("1 second(s)"))  : "1 second(s)"
   Loc: [/home/xxx/src/kf5/kcoreaddons/autotests/kformattest.cpp(318)]
PASS   : KFormatTest::formatRelativeDate()
PASS   : KFormatTest::formatValue()
PASS   : KFormatTest::cleanupTestCase()
Totals: 6 passed, 2 failed, 0 skipped, 0 blacklisted, 1ms
* Finished testing of KFormatTest *
  
  with installed kcoreaddons-lang
  
~/src/kf5/kcoreaddons-build> strace -e trace=file bin/kformattest  2>&1 | 
grep \.qm
access("/usr/share/locale/en/LC_MESSAGES/kcoreaddons5_qt.qm", R_OK) = 0
open("/usr/share/locale/en/LC_MESSAGES/kcoreaddons5_qt.qm", 
O_RDONLY|O_CLOEXEC) = 5
access("/usr/share/locale/de/LC_MESSAGES/kcoreaddons5_qt.qm", R_OK) = 0
open("/usr/share/locale/de/LC_MESSAGES/kcoreaddons5_qt.qm", 
O_RDONLY|O_CLOEXEC) = 5
access("/home/xxx/src/kf5/kcoreaddons/kcoreaddons5_qt.qm", R_OK) = 0
stat("/home/xxx/src/kf5/kcoreaddons/kcoreaddons5_qt.qm", 
{st_mode=S_IFREG|0644, st_size=729, ...}) = 0
open("/home/xxx/src/kf5/kcoreaddons/kcoreaddons5_qt.qm", 
O_RDONLY|O_CLOEXEC) = 5

```~/src/kf5/kcoreaddons-build> bin/kformattest 
* Start testing of KFormatTest *
Config: Using QtTest library 5.11.1, Qt 5.11.1 (x86_64-little_endian-lp64 
shared (dynamic) release build; by GCC 4.8.5)
PASS   : KFormatTest::initTestCase()
PASS   : KFormatTest::formatByteSize()
FAIL!  : KFormatTest::formatDuration() Compared values are not the same
   Actual   (format.formatDuration(singleSecond, options)): "0 h 00 m 04 s"
   Expected (QStringLiteral("0h00m04s"))  : "0h00m04s"
   Loc: [/home/xxx/src/kf5/kcoreaddons/autotests/kformattest.cpp(226)]
FAIL!  : KFormatTest::formatDecimalDuration() Compared values are not the 
same
   Actual   (format.formatDecimalDuration(10))   : "10 milliseconds"
   Expected (QStringLiteral("10 millisecond(s)")): "10 millisecond(s)"
   Loc: [/home/xxx/src/kf5/kcoreaddons/autotests/kformattest.cpp(302)]
FAIL!  : KFormatTest::formatSpelloutDuration() Compared values are not the 
same
   Actual   (format.formatSpelloutDuration(1000)): "1 second"
   Expected (QStringLiteral("1 second(s)"))  : "1 second(s)"
   Loc: [/home/xxx/src/kf5/kcoreaddons/autotests/kformattest.cpp(318)]
FAIL!  : KFormatTest::formatRelativeDate() Compared values are not the same
   Actual   (format.formatRelativeDate(testDate, QLocale::LongFormat)): 
"Heute"
   Expected (QStringLiteral("Today")) : 
"Today"
   Loc: [/home/xxx/src/kf5/kcoreaddons/autotests/kformattest.cpp(341)]
PASS   : KFormatTest::formatValue()
PASS   : KFormatTest::cleanupTestCase()
Totals: 4 passed, 4 failed, 0 skipped, 0 blacklisted, 1ms
* Finished testing of KFormatTest *

 

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  In D14779#308603 , @aacid wrote:
  
  > I think that in the test we just need to call
  >
  >   bool QCoreApplication::installTranslator(QTranslator *translationFile)
  >
  > with the location of the fake translation
  
  
  
  
  > and since newest location is tried first it should overwrite the one from 
the ecm loader.
  
  According to 
http://code.qt.io/cgit/qt/qtbase.git/tree/src/corelib/kernel/qcoreapplication.cpp#n1985
 new translators are prepended - not overwritten. Not sure if a missing 
translation in the latest file added would be search in the previous 
translators.
  
  To  understand you correctly: You really are thinking about generating  a qm 
file  from the kcoreaddons lib source with gettext containing real translations 
 added with lokalize or similar to avoid this issue ?

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks
Cc: aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309388 , @habacker wrote:
  
  > Agreed, as long as there is Qt on Windows not patched to also use unix 
style path layout, which is the case on OBS
  
  
  but this case should be covered by ECM module 'KDEInstallDirs'.
  
  If KDE_INSTALL_XXX always return a relative dir , it should be save to use

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309370 , @bcooksley 
wrote:
  
  > On Windows the correct path to use is $prefix/bin/data/... as that is what 
QStandardPaths expects.
  >  The location $prefix/share/ won't work on Windows, so the test will need 
to handle that case.
  
  
  Agreed, as long as there is Qt on Windows not patched to also use unix style 
path layout, which is the case on OBS

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  another option would be to force the install path in the cmake call
  
add_test(kdoctools_install ${CMAKE_CTEST_COMMAND}
  --build-and-test
  "${CMAKE_CURRENT_SOURCE_DIR}/kdoctools-install"
  "${CMAKE_CURRENT_BINARY_DIR}/kdoctools-install"
  --build-generator ${CMAKE_GENERATOR}
  --build-makeprogram ${CMAKE_MAKE_PROGRAM}
  --build-target install
  --build-options
  "-DCMAKE_MODULE_PATH=${CMAKE_MODULE_PATH}"
  
"-DCMAKE_INSTALL_PREFIX:PATH=${CMAKE_CURRENT_BINARY_DIR}/kdoctools-install/destdir"
  "-DKF5DocTools_DIR=${CMAKE_BINARY_DIR}/cmake"
  "-DKDOCTOOLS_CUSTOMIZATION_DIR=${KDOCTOOLS_CUSTOMIZATION_DIR}"
  "-DKDOCTOOLS_ENABLE_HTMLHANDBOOK=1"
 ...
"-DKDE_INSTALL_MANDIR=share/man"
"-DKDE_INSTALL_DOCBUNDLEDIR=share/doc/HTML"

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  Because test.qmake do not anything about KDE  install path layout those 
pathes need to be specified in the ctest config in autotests/CMakeLists.txt I 
guess
  
--test-command ${CMAKE_COMMAND}
"-DKDE_INSTALL_MANDIR=${KDE_INSTALL_MANDIR}"
"-DKDE_INSTALL_DOCBUNDLEDIR=${KDE_INSTALL_DOCBUNDLEDIR}"

"-DCMAKE_INSTALL_PREFIX:PATH=${CMAKE_CURRENT_BINARY_DIR}/kdoctools-install/destdir"
-P "${CMAKE_CURRENT_SOURCE_DIR}/kdoctools-install/test.cmake")
  
  and in test.qmake
  
...
set(MANPATH ${CMAKE_INSTALL_PREFIX}/${KDE_INSTALL_MANDIR})
set(HTMLPATH ${CMAKE_INSTALL_PREFIX}/${KDE_INSTALL_DOCBUNDLEDIR})
 ...

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309332 , @bcooksley 
wrote:
  
  > The Windows failure looks like a problem with handling of DESTDIR - 
probably due to reliance on it being handled by $DESTDIR instead of %DESTDIR% 
(which is the correct syntax for Windows)
  
  
  No , it is a install path mismatch.  According to 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20WindowsMSVCQt5.10/39/console
 the files are installed below <...>/bin/data 
(https://cgit.kde.org/extra-cmake-modules.git/tree/kde-modules/KDEInstallDirs.cmake)
 but test.qmake expects them in <...>/share 
(https://cgit.kde.org/kdoctools.git/tree/autotests/kdoctools-install/test.cmake)

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309303 , @bcooksley 
wrote:
  
  > Try removing that and the error it has should show up on your local system
  
  
  I see. 
  Since this problem has been fixed in kdoctools git repo and kdoctools has 
been rebuilt on KDE CI, packages like kio, which depend on kdoctools, are 
automatically rebuilt, as is the case with OBS?

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309295 , @habacker wrote:
  
  > According to https://cgit.kde.org/kdoctools.git/tree/src/meinproc.cpp#n51 
uses meinproc5  qCCritical(KDocToolsLog) - How can this be enabled on KDE CI ?
  
  
  Got it - see 
https://phabricator.kde.org/R238:81a3d951169b7f575c0a09c2cf19eb911fab9882

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309281 , @bcooksley 
wrote:
  
  > Dependency Builds triggered to roll it out to all the non Frameworks jobs 
which have already picked this up.
  
  
  How to do this ?

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  According to https://cgit.kde.org/kdoctools.git/tree/src/meinproc.cpp#n51 
uses meinproc5  qCCritical(KDocToolsLog) - How can this be enabled on KDE CI ?

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309285 , @habacker wrote:
  
  > Funny, that this works with Qt 5.9 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20SUSEQt5.9/ and Qt 
5.11 (locally tested) - it only happens with Qt 5.10
  
  
  Because I cannot reproduce this issue on a local build there is the question 
how

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309283 , @habacker wrote:
  
  > This is the remaining issue on non Windows KDE CI Qt 5.10
  >
  > 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20SUSEQt5.10/49/consoleFull
  >
  >   07:02:43 cd "/home/jenkins/workspace/Frameworks kdoctools kf5-qt5 
SUSEQt5.10/autotests/kdoctools-install" && "/home/jenkins/workspace/Frameworks 
kdoctools kf5-qt5 SUSEQt5.10/build/bin/meinproc5" --check --cache 
/home/jenkins/workspace/Frameworks\ kdoctools\ kf5-  qt5\ 
SUSEQt5.10/build/autotests/kdoctools-install/po/es/docs/foobar/index.cache.bz2 
po/es/docs/foobar/index.docbook
  >   07:02:43 gmake[2]: *** 
[CMakeFiles/po-es-docs-foobar-index-cache-bz2.dir/build.make:65: 
po/es/docs/foobar/index.cache.bz2] Error 1
  
  
  Funny, that this works with Qt 5.9 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20SUSEQt5.9/ and Qt 
5.11 (locally tested) - it only happens with Qt 5.10

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#309281 , @bcooksley 
wrote:
  
  > This change has caused CI wide breakage and makes kdoctools unrelocatable.
  
  
  This has been fixed with commit  
https://phabricator.kde.org/R238:4e48476f47414f1077a181ef56a19f7d9289d50d
  
  See other comments for the remaining issues.

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  This is the remaining issue on non Windows KDE CI Qt 5.10
  
  
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20SUSEQt5.10/49/consoleFull
  
07:02:43 cd "/home/jenkins/workspace/Frameworks kdoctools kf5-qt5 
SUSEQt5.10/autotests/kdoctools-install" && "/home/jenkins/workspace/Frameworks 
kdoctools kf5-qt5 SUSEQt5.10/build/bin/meinproc5" --check --cache 
/home/jenkins/workspace/Frameworks\ kdoctools\ kf5-  qt5\ 
SUSEQt5.10/build/autotests/kdoctools-install/po/es/docs/foobar/index.cache.bz2 
po/es/docs/foobar/index.docbook
07:02:43 gmake[2]: *** 
[CMakeFiles/po-es-docs-foobar-index-cache-bz2.dir/build.make:65: 
po/es/docs/foobar/index.cache.bz2] Error 1

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: bcooksley, kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, 
ngraham, bruns, skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  I applied a fix to use absolute pathes. Unfortunally windows builds have an 
additional issue:
  
  > 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20WindowsMSVCQt5.10/36/console
  
21:26:34 -- Installing: C:/CI/workspace/Frameworks kdoctools kf5-qt5 
WindowsMSVCQt5.10/build/autotests/kdoctools-install/destdir/bin/data/man/man1/checkXML5.1
...
21:26:34   C:/CI/workspace/Frameworks kdoctools kf5-qt5 
WindowsMSVCQt5.10/build/autotests/kdoctools-install/destdir/share/man/man1/checkXML5.1
  was not found
  
  -> msvc builds uses a different install dir 
  This is caused by 
https://cgit.kde.org/extra-cmake-modules.git/tree/kde-modules/KDEInstallDirs.cmake#n443

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, ngraham, bruns, 
skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

habacker added a comment.


  In D14817#308903 , @ltoscano wrote:
  
  > This worked with Qt 5.9 on the CI and locally with 5.10 too, but the CI 
seems to be unhappy for openSUSE/Qt 5.10 and Windows/Qt 5.10 (it works on 
openSUSE/Qt 5.9 and FreeBSD/Qt 5.10):
  >
  > - 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20SUSEQt5.10/46/
  
  
  14:08:13 Scanning dependencies of target po-es-docs-foobar-index-cache-bz2
  14:08:13 [  8%] Generating po/es/docs/foobar/index.cache.bz2
  14:08:13 gmake[2]: *** 
[CMakeFiles/po-es-docs-foobar-index-cache-bz2.dir/build.make:62: 
po/es/docs/foobar/index.cache.bz2] Error 1
  14:08:13 gmake[1]: *** [CMakeFiles/Makefile2:73: 
CMakeFiles/po-es-docs-foobar-index-cache-bz2.dir/all] Error 2
  14:08:13 gmake: *** [Makefile:130: all] Error 2
  reason not visible - I suggest to add VERBOSE=1 to the related make comment
  
  > - 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20WindowsMSVCQt5.10/35/
  
  14:11:43 Test command failed: 
C:/Craft/CI/windows-msvc2017_64-cl-debug/dev-utils/cmake/bin/cmake.exe
  14:11:43 CMake Error at C:/CI/workspace/Frameworks kdoctools kf5-qt5 
WindowsMSVCQt5.10/autotests/kdoctools-install/test.cmake:20 (message):
  14:11:43   destdir/share/man/man1/checkXML5.1 was not found
  This may be caused  by not specifing an absolute path in test.cmake e.g.
  
-set(MANPATH destdir/share/man)
-set(HTMLPATH destdir/share/doc/HTML)
+set(MANPATH ${CMAKE_INSTALL_PREFIX}/destdir/share/man)
+set(HTMLPATH ${CMAKE_INSTALL_PREFIX}/destdir/share/doc/HTML)

---test-command ${CMAKE_COMMAND}  -P 
"${CMAKE_CURRENT_SOURCE_DIR}/kdoctools-install/test.cmake
+--test-command ${CMAKE_COMMAND}  
"-DCMAKE_INSTALL_PREFIX:PATH=${CMAKE_CURRENT_BINARY_DIR}/kdoctools-install/destdir"
 -P "${CMAKE_CURRENT_SOURCE_DIR}/kdoctools-install/test.cmake"
  
  > - 
https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20FreeBSDQt5.10/
  
  #1​6 Aug 14, 2018 2:07 PM
  Finished: SUCCESS 
  no issue visible -> tests are disabled
  
  > - https://build.kde.org/job/Frameworks%20kdoctools%20kf5-qt5%20SUSEQt5.9/32/
  
  #3​2 Aug 14, 2018 2:07 PM
  Finished: SUCCESS 
  no issue visible -> tests are disabled

REPOSITORY
  R238 KDocTools

REVISION DETAIL
  https://phabricator.kde.org/D14817

To: habacker, ltoscano
Cc: kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, ngraham, bruns, 
skadinna

D14817: Convert manual install test to an autotest

[Ralf Habacker]

This revision was automatically updated to reflect the committed changes.
Closed by commit R238:f0cc8b2ccb4a: Convert manual install test to an autotest 
(authored by habacker).
Herald added a project: Frameworks.
Herald added a subscriber: kde-frameworks-devel.

CHANGED PRIOR TO COMMIT
  https://phabricator.kde.org/D14817?vs=39659=39697#toc

REPOSITORY
  R238 KDocTools

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D14817?vs=39659=39697

REVISION DETAIL
  https://phabricator.kde.org/D14817

AFFECTED FILES
  CMakeLists.txt
  KF5DocToolsConfig.cmake.in
  autotests/CMakeLists.txt
  autotests/kdoctools-install/CMakeLists.txt
  autotests/kdoctools-install/po/es/docs/foobar/index.docbook
  autotests/kdoctools-install/po/es/docs/kioslave5/fooslave/index.docbook
  autotests/kdoctools-install/po/es/docs/man-kjscmd.1.docbook
  autotests/kdoctools-install/po/fr/docs/foobar/index.docbook
  autotests/kdoctools-install/po/fr/docs/man-kjscmd.1.docbook
  autotests/kdoctools-install/test.cmake
  tests/kdoctools_install-test/po/es/docs/foobar/index.docbook
  tests/kdoctools_install-test/po/es/docs/kioslave5/fooslave/index.docbook
  tests/kdoctools_install-test/po/es/docs/man-kjscmd.1.docbook
  tests/kdoctools_install-test/po/fr/docs/foobar/index.docbook
  tests/kdoctools_install-test/po/fr/docs/man-kjscmd.1.docbook

To: habacker, ltoscano
Cc: kde-frameworks-devel, ltoscano, kde-doc-english, michaelh, ngraham, bruns, 
skadinna

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker added a comment.


  You have remembered that in this case the loader created by 
ecm_create_qm_loader(kcoreaddons_QM_LOADER kcoreaddons5_qt) must be adjusted to 
find this - not installed - fake translation by default ?

REPOSITORY
  R244 KCoreAddons

REVISION DETAIL
  https://phabricator.kde.org/D14779

To: habacker, #frameworks, cfeck
Cc: aacid, kde-frameworks-devel, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker updated this revision to Diff 39684.
habacker edited the summary of this revision.
habacker added a comment.


  - after feedback from ecm maintainers the fix could be refactored to not be 
depending on ecm patch

REPOSITORY
  R244 KCoreAddons

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D14779?vs=39560=39684

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14779

AFFECTED FILES
  autotests/kformattest.cpp

To: habacker, #frameworks, cfeck
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14778: Disable loading of translations when using 'C' locale

[Ralf Habacker]

habacker abandoned this revision.
habacker added a comment.


  How to handle single/plural form then ?
  
  QString KFormatPrivate::formatDecimalDuration(quint64 msecs, int 
decimalPlaces) const
  {
  ...
  
//: @item:intext %1 is a whole number
//~ singular %n millisecond
//~ plural %n milliseconds
return tr("%n millisecond(s)", nullptr, msecs);
  
  According to http://doc.qt.io/qt-5/i18n-source-translation.html the recent 
string usage seems to be exactly what is required.
  
  From further inspection the solutions is to use translated reference values
  
  - QCOMPARE(format.formatDecimalDuration(10), QStringLiteral("10 
millisecond(s)"));
  - QCOMPARE(format.formatDecimalDuration(10, 3), QStringLiteral("10 
millisecond(s)"));
  
  +QCOMPARE(format.formatDecimalDuration(10), 
QCoreApplication::translate("KFormat", "%n millisecond(s)", nullptr, 10));
  +QCOMPARE(format.formatDecimalDuration(10, 3), 
QCoreApplication::translate("KFormat", "%n millisecond(s)", nullptr, 10));
  
  Thanks for your feedback.

REPOSITORY
  R240 Extra CMake Modules

REVISION DETAIL
  https://phabricator.kde.org/D14778

To: habacker, #frameworks, apol
Cc: aacid, kde-frameworks-devel, kde-buildsystem, michaelh, ngraham, bruns

D14787: Fix 'ThemeTest::testCompositingChange() fails if windows manager is running'

[Ralf Habacker]

habacker created this revision.
Restricted Application added a project: Frameworks.
Restricted Application added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  Skip the test in that case and show the reason.
  
  BUG: 397418
  FIXED-IN: 5.50.0

TEST PLAN
  compiled and test on opensuse os

REPOSITORY
  R242 Plasma Framework (Library)

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14787

AFFECTED FILES
  autotests/themetest.cpp

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14785: Fix DialogNativeTest in multi screen configuration

[Ralf Habacker]

habacker created this revision.
Restricted Application added a project: Frameworks.
Restricted Application added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  On multi screen configuration the x coordinate of the primary screen may
  not be always zero.
  
  With this commit, the widgets are displayed in different colors to
  distinguish them better.
  
  BUG: 397417
  FIXED-IN: 5.50.0

REPOSITORY
  R242 Plasma Framework (Library)

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14785

AFFECTED FILES
  autotests/dialognativetest.cpp

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14778: Disable loading of translations when using 'C' locale

[Ralf Habacker]

habacker added a comment.


  Just as info:
  
  kcoreaddons uses '(s)' suffix in its internal formating  methods 
(https://cgit.kde.org/kcoreaddons.git/commit/src/lib/util/kformatprivate.cpp?id=ffcd094c1f4716a0d91650608e42017886a0a7e3)
 which is always translated with or without 's'  'not always (s)' depending on 
the amount (see 
https://websvn.kde.org/trunk/l10n-kf5/en/messages/frameworks/kcoreaddons5_qt.po?revision=1522490=markup
 line 440ff)

REPOSITORY
  R240 Extra CMake Modules

REVISION DETAIL
  https://phabricator.kde.org/D14778

To: habacker, #frameworks, apol
Cc: aacid, kde-frameworks-devel, kde-buildsystem, michaelh, ngraham, bruns

D14778: Disable loading of translations when using 'C' locale

[Ralf Habacker]

habacker added a comment.


  No, both LC_ALL=C and LC_ALL=en  fetches the same english translation
  
~/src/kf5/kcoreaddons-build> LC_ALL=en strace -e trace=file bin/kformattest 
2>&1 | grep \.qm | grep R_OK
access("/usr/share/locale/en/LC_MESSAGES/kcoreaddons5_qt.qm", R_OK) = 0

~/src/kf5/kcoreaddons-build> LC_ALL=C strace -e trace=file bin/kformattest 
2>&1 | grep \.qm | grep R_OK
access("/usr/share/locale/en/LC_MESSAGES/kcoreaddons5_qt.qm", R_OK) = 0
  
  The result is shown at https://bugs.kde.org/show_bug.cgi?id=397404#c0 (see 
below !!and with 'C' locale!!)

REPOSITORY
  R240 Extra CMake Modules

REVISION DETAIL
  https://phabricator.kde.org/D14778

To: habacker, #frameworks, apol
Cc: aacid, kde-frameworks-devel, kde-buildsystem, michaelh, ngraham, bruns

D14779: Fix 'kformattest fails with installed kcoreaddons language package'

[Ralf Habacker]

habacker created this revision.
Restricted Application added a project: Frameworks.
Restricted Application added a subscriber: kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  See D14778  for the related 
extra-cmake-modules patch
  
  BUG:397404
  FIXED-IN:5.50.0

REPOSITORY
  R244 KCoreAddons

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14779

AFFECTED FILES
  autotests/kformattest.cpp

To: habacker
Cc: kde-frameworks-devel, michaelh, ngraham, bruns

D14778: Disable loading of translations when using 'C' locale

[Ralf Habacker]

habacker created this revision.
Restricted Application added projects: Frameworks, Build System.
Restricted Application added subscribers: kde-buildsystem, kde-frameworks-devel.
habacker requested review of this revision.

REVISION SUMMARY
  kcoreaddons test case 'kformattest' is designed not to use translations
  and fails if a translation of kcoreaddons is installed.
  To disable translations, QLocale::setDefault(QLocale::C) can now be called 
before
  instantiating Q[Core]Application.
  
  CCBUG:397404

REPOSITORY
  R240 Extra CMake Modules

BRANCH
  master

REVISION DETAIL
  https://phabricator.kde.org/D14778

AFFECTED FILES
  modules/ECMQmLoader.cpp.in

To: habacker
Cc: kde-frameworks-devel, kde-buildsystem, michaelh, ngraham, bruns

D9387: Add cross compile support for kconfig_compiler and kconf_update.

[Ralf Habacker]

habacker added a comment.


  Sorry for the two spam comments, I had a freeze at phabricator web site.
  
  > Here's the android scripts for reference. 
https://cgit.kde.org/sysadmin/ci-tooling.git/tree/system-images/android/sdk
  
  Ah I see, on Android the packages containing native tools are build. The 
package and their development packages including the cmake support files 
(xxxTargets.cmake)  are then installed to a custom destination  to which 
KF5_HOST_TOOLING is pointing to. Transformed to obs that would mean to use the 
runtime and development packages from 
https://build.opensuse.org/project/show/KDE:Frameworks5.
  
  From working on the kf5 packages on obs I know that this approach has some 
issues:
  
  1. Not all required tools are installed with the package - I needed dedicated 
patches to let packages install them too (at least)
- 
https://build.opensuse.org/package/show/home:rhabacker:branches:KDE:Frameworks5/kdoctools-17
- 
https://build.opensuse.org/package/show/home:rhabacker:branches:KDE:Frameworks5/sonnet-17
  2. From memory  I remember that not  all KF5 packages have working cross 
compile support
  3. obs uses currently another approach - there is a cross kf5 tools package 
providing all required tools installed in a custom location in one package 
(https://build.opensuse.org/package/view_file/home:rhabacker:branches:windows:mingw:win32:KF547/mingw32-cross-kf5-tools/mingw32-cross-kf5-tools.spec?expand=1).
 This behavior is derived from the related xxx-cross-libqt5-yyy package. The 
native tools are installed as mingw helper tools and used in the cmake command 
line 
https://build.opensuse.org/package/view_file/home:rhabacker:branches:windows:mingw:win32:KF547/mingw32-kf5-filesystem/macros.mingw32-kf5?expand=1

REPOSITORY
  R237 KConfig

REVISION DETAIL
  https://phabricator.kde.org/D9387

To: habacker, dfaure, ltoscano, bcooksley, apol
Cc: kde-frameworks-devel, apol, michaelh, ngraham, bruns

D9387: Add cross compile support for kconfig_compiler and kconf_update.

[Ralf Habacker]

habacker added a comment.


  
  
  In D9387#180530 , @apol wrote:
  
  > Yes, defining KF5_HOST_TOOLING should be enough, kconf_update will need to 
be exported with kconfigcompiler though, at the moment it's not.
  
  
  that means host tooling support in kconfig is not usable in kconfig yet

REPOSITORY
  R237 KConfig

REVISION DETAIL
  https://phabricator.kde.org/D9387

To: habacker, dfaure, ltoscano, bcooksley, apol
Cc: kde-frameworks-devel, apol, michaelh, ngraham, bruns

D9387: Add cross compile support for kconfig_compiler and kconf_update.

[Ralf Habacker]

habacker added a comment.


  The KF5_HOST_TOOLING approach in kconfig does not work because it is not poss
  
  In D9387#180530 , @apol wrote:
  
  > Yes, defining KF5_HOST_TOOLING should be enough, kconf_update will need to 
be exported with kconfigcompiler though, at the moment it's not.
  
  
  so you are saying that host tooling support  in kconfig does not work yet
  
  > Here's the android scripts for reference. 
https://cgit.kde.org/sysadmin/ci-tooling.git/tree/system-images/android/sdk
  
  But this link is useless because it does not show  how  the included file 
cmake file should be setup up.

REPOSITORY
  R237 KConfig

REVISION DETAIL
  https://phabricator.kde.org/D9387

To: habacker, dfaure, ltoscano, bcooksley, apol
Cc: kde-frameworks-devel, apol, michaelh, ngraham, bruns

D9387: Add cross compile support for kconfig_compiler and kconf_update.

[Ralf Habacker]

habacker added a comment.


  The KF5_HOST_TOOLING approach in kconfig does not work because it is not poss
  
  In D9387#180530 , @apol wrote:
  
  > Yes, defining KF5_HOST_TOOLING should be enough, kconf_update will need to 
be exported with kconfigcompiler though, at the moment it's not.
  
  
  you are saying that host tooling support  in kconfig does not work yet

REPOSITORY
  R237 KConfig

REVISION DETAIL
  https://phabricator.kde.org/D9387

To: habacker, dfaure, ltoscano, bcooksley, apol
Cc: kde-frameworks-devel, apol, michaelh, ngraham, bruns

D9387: Add cross compile support for kconfig_compiler and kconf_update.

[Ralf Habacker]

habacker added a comment.


  In D9387#305685 , @dfaure wrote:
  
  >
  
  
  
  
  > I don't know much about all this, but it sounds like Aleix is saying that 
an existing solution also works and is simpler.
  
  Please show me that solution with a working example

REPOSITORY
  R237 KConfig

REVISION DETAIL
  https://phabricator.kde.org/D9387

To: habacker, dfaure, ltoscano, bcooksley, apol
Cc: kde-frameworks-devel, apol, michaelh, ngraham, bruns

D9387: Add cross compile support for kconfig_compiler and kconf_update.

[Ralf Habacker]

habacker added a comment.


  This patch is used at
  
  
https://build.opensuse.org/package/show/home:rhabacker:branches:windows:mingw:win32:KF547/mingw32-kconfig
  
https://build.opensuse.org/package/show/home:rhabacker:branches:windows:mingw:win64:KF547/mingw64-kconfig
  
  for about a year without any problems

REPOSITORY
  R237 KConfig

REVISION DETAIL
  https://phabricator.kde.org/D9387

To: habacker, dfaure, ltoscano, bcooksley, apol
Cc: kde-frameworks-devel, apol, michaelh, ngraham, bruns

D13743: Migrate build system to use find_package in autotests/ki18n_install

[Ralf Habacker]

This revision was automatically updated to reflect the committed changes.
Closed by commit R249:507415c54bd1: Migrate build system to use find_package in 
autotests/ki18n_install (authored by habacker).

CHANGED PRIOR TO COMMIT
  https://phabricator.kde.org/D13743?vs=38569=39339#toc

REPOSITORY
  R249 KI18n

CHANGES SINCE LAST UPDATE
  https://phabricator.kde.org/D13743?vs=38569=39339

REVISION DETAIL
  https://phabricator.kde.org/D13743

AFFECTED FILES
  CMakeLists.txt
  autotests/CMakeLists.txt
  autotests/ki18n_install/CMakeLists.txt

To: habacker, apol, ilic, dfaure
Cc: dfaure, bcooksley, ltoscano, kde-frameworks-devel, michaelh, ngraham, bruns