[Kmymoney-devel] Problem building KMM using clang on Mac OS X

Tue Jan 28 11:36:40 UTC 2014

On 28/01/14 10:44, Allan wrote:
> On 28/01/14 08:01, Cristian Oneț wrote:
>> Allan,
>>
>> Nothing has changed in kdeui that's causing this, the problem is
>> actually the following:
>> The last commit to csvutil.cpp introduced a dependency on kdeui. The
>> parsedatatest_SRCS set includes csvutil.cpp but when linking
>> ${KDE4_KDEUI_LIBS} is missing. You should revisit the tests to see if
>> they are still valid considering the fact that they have a dependency
>> on kdeui. In the meantime add ${KDE4_KDEUI_LIBS} to the
>> TARGET_LINK_LIBRARIES of parsedatatest.
>>
>> Regards,
>> Cristian
>>
>> [1]
>> https://projects.kde.org/projects/extragear/office/kmymoney/repository/revisions/master/entry/kmymoney/plugins/csvimport/CMakeLists.txt
>>
>>
>
> Hi Cristian
>
> Yes, I've already done that to produce the patch I was referring to in
> my earlier post.
>
> In fact, then the following test also showed the same problem, so both
> cmake entries needed the fix.
>
> I'm intending to follow up to find what actually caused the dependency.
>   In fact, I'd moved quite a chunk of code into csvutil, from
> InvestProcessing, so that CSVDialog could access it, but at the moment I
> can't think of anything relating to the tests.
>
> The plugin itself already was linked to ${KDE4_KDEUI_LIBS}, and the two
> tests were linked to csvutil.  Might that be the answer?
>
> Allan

The problem in not in the tests at all, but in CsvUtil.  I'd moved into 
CsvUtil from InvestProcessing the function CsvUtil::createAccount(), 
which I'd originally borrowed/adapted from the QIF importer (I think), 
and it contains references to KMessageBox::questionYesNoCancel(0, 
message) and KMessageBox::information(0, i18n("Unable to add account: 
%1", e->what())).  If I 'no-op'ed them both, the build was successful, 
without editing CMakeLists.txt.  The two tests do not reference 
CsvUtil::createAccount(), and have not been changed for a long while.

So, I'll return things to normal, then test the patch and do a commit.

Allan

>> 2014-01-28 Allan <agander93 at gmail.com>:
>>> On 27/01/14 20:28, Alvaro Soliverez wrote:
>>>>
>>>> According to Jenkings, the last commit failed while building the cvs
>>>> import code.
>>>>
>>>> http://build.kde.org/job/kmymoney_master/188/
>>>
>>>
>>> Never heard of him before!  Was that something that you received
>>> notification of, or did you go look?
>>>
>>> Either way, that output was a bit more helpful.  I've run the build
>>> tests
>>> several times over the last few days.  I even made a post about an error
>>> elsewhere "Unit Test #28: kmymoney-querytabletest Fails" on 24 Jan.
>>>
>>> However, when I tried again, I got the failure too.  I had to make the
>>> suggested change to two different tests, neither of which has been
>>> changed
>>> for ages.  Does that suggest that libkdeui.so.5 has changed? But
>>> then, I've
>>> not done a system update for several days, and the file is dated July
>>> 2013.
>>> So, I'm mystified.  My last commit did make a small and unconnected
>>> change
>>> to csvutil.  Might that have triggered something?
>>>
>>> So, now I have a tiny patch.  Am I OK to push it?
>>>
>>> Allan
>>>
>>>
>>>>
>>>> On Mon, Jan 27, 2014 at 5:16 PM, Allan <agander93 at gmail.com> wrote:
>>>>>
>>>>> On 27/01/14 16:13, mk-lists at email.de wrote:
>>>>>>
>>>>>>
>>>>>> I was trying to build the latest git version of KMM on OSX and
>>>>>> experienced
>>>>>> a problem.
>>>>>> KDE is the current version:
>>>>>> ---
>>>>>> $ port installed kdelibs4
>>>>>> The following ports are currently installed:
>>>>>>      kdelibs4 @4.11.5_0+docs (active)
>>>>>> —
>>>>>>
>>>>>> KMessageBox::information() seems to be unknown:
>>>>>> ---
>>>>>> :info:build /usr/bin/clang++   -pipe -Os -arch x86_64  -fno-common
>>>>>> -Wnon-virtual-dtor -Wno-long-long -Wundef -Wcast-align
>>>>>> -Wchar-subscripts
>>>>>> -Wall -W -Wpointer-arith -Wformat-security -Woverloaded-virtual
>>>>>> -fno-common
>>>>>> -fvisibility==
>>>>>> hidden -Werror=return-type -fvisibility-inlines-hidden
>>>>>> -Wno-return-type-c-linkage -fexceptions -UQT_NO_EXCEPTIONS
>>>>>> -fexceptions
>>>>>> -UQT_NO_EXCEPTIONS -fPIC -fno-common -O2 -DNDEBUG -DQT_NO_DEBUG -arch
>>>>>> x86_64
>>>>>> -Wl,-search_paths_first --
>>>>>> Wl,-headerpad_max_install_names  -Wl,-headerpad_max_install_names
>>>>>> -arch
>>>>>> x86_64 CMakeFiles/parsedatatest.dir/parsedatatest_automoc.cpp.o
>>>>>> CMakeFiles/parsedatatest.dir/parsedatatest.cpp.o
>>>>>> CMakeFiles/parsedatatest.dir/csvutil.cpp.o
>>>>>> -o parsedatatest.app/Contents/MacOS/parsedatatest
>>>>>> /opt/local/lib/libQtTest.dylib /opt/local/lib/libkdecore.5.11.5.dylib
>>>>>> /opt/local/lib/libgmp.dylib /opt/local/lib/libalkimia.dylib
>>>>>> ../../../lib/libkmm_mymoney.4.11.5.dylib /opt/loo
>>>>>> cal/lib/libkdecore.5.11.5.dylib /opt/local/lib/libQtDBus.dylib
>>>>>
>>>>>
>>>>>
>>>>> I'm not qualified to give technical assistance in this area, but
>>>>> the last
>>>>> two lines above look strange, although I suspect it may be the
>>>>> result of
>>>>> a
>>>>> copy/paste operation?
>>>>>
>>>>> One other point is that there is reference to 'parsedatatest' in
>>>>> csvutil.
>>>>> Whether that is anything to do with the actual problem, I don't
>>>>> know, but
>>>>> I'd have been inclined to turn off the unit tests initially.  That
>>>>> test
>>>>> is
>>>>> very small and quite simplistic.
>>>>>
>>>>> That's my pennyworth.
>>>>>
>>>>> Allan
>>>
>>>
>>> _______________________________________________
>>> KMyMoney-devel mailing list
>>> KMyMoney-devel at kde.org
>>> https://mail.kde.org/mailman/listinfo/kmymoney-devel
>> _______________________________________________
>> KMyMoney-devel mailing list
>> KMyMoney-devel at kde.org
>> https://mail.kde.org/mailman/listinfo/kmymoney-devel
>>
> _______________________________________________
> KMyMoney-devel mailing list
> KMyMoney-devel at kde.org
> https://mail.kde.org/mailman/listinfo/kmymoney-devel