Trouble compiling KMM 5.0 and Master Branch on new Fedora 29 system

[Jack Ostroff]

On 2018.11.12 16:27, Jack wrote:
> OK, I've now been able to duplicate it - it does NOT happen if the  
> build directory is not under the source directory.  It does happen if  
> the build directory is under the source directory.  I still haven't  
> figured out exactly what is failing, but I finally do see the  
> relevant files left under CMakeFiles/CMakeTmp.
Got it (sort of).  the relevant error occurs when cmake runs gmake  
(symlink from make) on the Makefile it creates to compile the cxx  
program which tests for the presence of libIcal.

gmake[2]: Entering directory  
'/local/data/jack/KDE/KMM/kmymoney-MASTER-2018.11.12-13:15:37/build/CMakeFiles/CMakeTmp'
CMakeFiles/cmTC_98fb0.dir/build.make:64: *** target pattern contains no  
'%'.  Stop.

A bit of googling says that error is perhaps not well worded, but does  
occur because of the colon in the path name in line 64 of that  
build.make file (buried under CMakeTmp.)  The colon confuses the make  
about what rule it is trying to deal with on that line.

Because your script has worked for years, I suspect either something  
may have changed in a recent version of make, or perhaps you have moved  
the build dir from outside to under the src dir?

> On 2018.11.12 15:24, Brendan Coupe wrote:
>> I figured it out. After trying many manual and script based builds I  
>> started to see a pattern. It only failed to build when it was in the  
>> directory that my script created with a date/time stamp. I've been  
>> using this script for 10+ years without a problem. The download  
>> directory names looked like this:
>> 
>> kmymoney-MASTER-2018.11.12-13:15:37
>> 
>> Simply changing the : to . solve the cmake problem.
>> 
>> kmymoney-MASTER-2018.11.12-13.15.37
>> 
>> Sometimes I really hate computers.
>> 
>> Thanks for all of your help Jack.
>> 
>> ----
>> Brendan Coupe
>> 
>> 
>> On Sun, Nov 11, 2018 at 4:53 PM Jack  
>> <ostroffjh at users.sourceforge.net>
>> wrote:
>> 
>> > On 2018.11.10 20:25, Brendan Coupe wrote:
>> > > I'm not sure I explained what I've tried well enough for you to
>> > > understand. While I normally use compile and install scripts  
>> that I
>> > > wrote a long time ago (used with cvs and svn) I tried compiling  
>> from
>> > > the official tar file for the recently released 5.0.2. It worked  
>> fine
>> > > on both my older upgraded (F28 to F29) and on my new clean F29
>> > > install.
>> > >
>> > > I think this means my systems are fine.
>> > >
>> > > I just manually downloaded both the master and 5.0 branches from  
>> the
>> > > KMM git web site. Both compile fine. When I run the git clone  
>> command
>> > > that is in my script and then run cmake, it fails just like I
>> > > mentioned in my original email. Except for how I download the  
>> source
>> > > code, I'm running the exact same steps that I use for the  
>> manually
>> > > downloaded zip versions of KMM (5.0.2 release, 5.0 branch &  
>> master
>> > > branch) and they all succeed.
>> > >
>> > > When I compare the size of the source code directories, all of  
>> the
>> > > zip files result in about 21k of source files. The git downloaded
>> > > folder is about 70k. I suspect git is the problem and it may not  
>> be
>> > > related to F29 since I started having problems compiling from  
>> source
>> > > several weeks before I upgraded to F29. There is a .git  
>> directory in
>> > > the git clone directory that is about 49k so it's probably the
>> > > difference.
>> > You could do a diff -r comparing the source dir from the zip/tar  
>> and
>> > the source downloaded from git.  The only difference (off the top  
>> of my
>> > head) should be the presence of the .git directory, as you  
>> suspect.  If
>> > there are other differences, see what they are.  Be sure that the  
>> zip
>> > and git are both at the same commit of the same branch, which I  
>> think
>> > you are already doing correctly.
>> >
>> > > The git command that worked until 4 to 6 weeks ago is:
>> > >
>> > > git clone git://anongit.kde.org/kmymoney $BRANCH $KMMDIR
>> > >
>> > > Is this correct?
>> > I don't think so.  I don't think (but am not absolutely certain)  
>> that
>> > you shouldn't specify the branch.  Do the git clone first  
>> (specifying
>> > the directory is ok) and then do "git checkout branch" to switch.
>> > Looking at "git clone --help" however, you could also do "git  
>> clone URL
>> > --branch $BRANCH" to not need to do the checkout.
>> > >
>> > I'm also still curious whether there is any hint of what went  
>> wrong in
>> > those files or folder after the cmake fails.
>> >
>> > In theory, the git and tarball/zip of the same branch/commit  
>> should be
>> > identical, so the issue may be either that the git clone does have  
>> some
>> > difference, or that cmake is possibly aware it is running in a git
>> > repository and doing something different.  (I think the latter is
>> > unlikely, but I'm grabbing at straws here.)
>> >
>> > Jack
>> > > ----
>> > > Brendan Coupe
>> > >
>> > >
>> > > On Sat, Nov 10, 2018 at 3:44 PM Jack  
>> <ostroffjh at users.sourceforge.net>
>> > > wrote:
>> > >
>> > > > On 2018.11.10 13:35, Brendan Coupe wrote:
>> > > > > I downloaded the source for 5.0.2 and compiled it without any
>> > > > > problems. I
>> > > > > tried the 5.0 and master branches again and they failed on  
>> iCal
>> > > again.
>> > > > > Seems that there is something off with the source on git???
>> > > > I doubt there is actually anything "wrong."  I suspect some bad
>> > > > interaction caused by F29 using a newer version of some tool.
>> > > Again -
>> > > > right after you get the failure, have you checked the files and
>> > > > directory I suggested?   I'll also try compiling from git head  
>> with
>> > > gcc
>> > > > 8.
>> > > >
>> > > > >
>> > > > > ----
>> > > > > Brendan Coupe
>> > > > >
>> > > > >
>> > > > > On Thu, Nov 8, 2018 at 11:44 AM Jack
>> > > <ostroffjh at users.sourceforge.net>
>> > > > > wrote:
>> > > > >
>> > > > > > I am able to run cmake (3.11.4) with gcc 8.2.0, with either
>> > > make or
>> > > > > > ninja, with no errors.   In your build directory, is there
>> > > anything
>> > > > > > useful in CMakeFiles/CMakeOutput.log or CMakeError.log?   
>> Also
>> > > look
>> > > > > in
>> > > > > > CMakeTmp/
>> > > > > >
>> > > > > > Another possible approach:   cmake --all-other-parameters
>> > > > > >  
>> --trace-source=/path/to/source/cmake/modules/FindLinical.cmake
>> > > > > > --debug-trycompile /path/to/source
>> > > > > > Then see if anything useful shows up in the above places.
>> > > > > >
>> > > > > > On 2018.11.07 20:29, Brendan Coupe wrote:
>> > > > > > > Fedora got 5.0.1 to compile so there must be some hope:-)
>> > > > > > >
>> > > > > > > I added the --debug-output option to cmake. The output  
>> from
>> > > just
>> > > > > after
>> > > > > > > the weboob errors are copied below:
>> > > > > > >
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> ===========================================================================
>> > > > > > >
>> > > > > > > CMake Warning (dev) at cmake/modules/FindLibical.cmake:59
>> > > > > (try_run):
>> > > > > > >   Policy CMP0066 is not set: Honor per-config flags in
>> > > > > try_compile()
>> > > > > > >   source-file signature.  Run "cmake --help-policy  
>> CMP0066"
>> > > for
>> > > > > policy
>> > > > > > >   details.  Use the cmake_policy command to set the  
>> policy and
>> > > > > > > suppress this
>> > > > > > >   warning.
>> > > > > > >
>> > > > > > >   For compatibility with older versions of CMake,  
>> try_compile
>> > > is
>> > > > > not
>> > > > > > > honoring
>> > > > > > >   caller config-specific compiler flags (e.g.
>> > > > > CMAKE_C_FLAGS_DEBUG)
>> > > > > > > in the
>> > > > > > >   test project.
>> > > > > > > Call Stack (most recent call first):
>> > > > > > >   CMakeLists.txt:242 (find_package)
>> > > > > > > This warning is for project developers.  Use -Wno-dev to
>> > > suppress
>> > > > > it.
>> > > > > > >
>> > > > > > >    Called from: [2]
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/cmake/modules/FindLibical.cmake
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > CMake Warning (dev) at cmake/modules/FindLibical.cmake:59
>> > > > > (try_run):
>> > > > > > >   Policy CMP0056 is not set: Honor link flags in  
>> try_compile()
>> > > > > > > source-file
>> > > > > > >   signature.  Run "cmake --help-policy CMP0056" for  
>> policy
>> > > > > details.
>> > > > > > > Use the
>> > > > > > >   cmake_policy command to set the policy and suppress  
>> this
>> > > > > warning.
>> > > > > > >
>> > > > > > >   For compatibility with older versions of CMake,  
>> try_compile
>> > > is
>> > > > > not
>> > > > > > > honoring
>> > > > > > >   caller link flags (e.g.  CMAKE_EXE_LINKER_FLAGS) in  
>> the test
>> > > > > > > project.
>> > > > > > > Call Stack (most recent call first):
>> > > > > > >   CMakeLists.txt:242 (find_package)
>> > > > > > > This warning is for project developers.  Use -Wno-dev to
>> > > suppress
>> > > > > it.
>> > > > > > >
>> > > > > > >    Called from: [2]
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/cmake/modules/FindLibical.cmake
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > CMake Warning (dev) at cmake/modules/FindLibical.cmake:59
>> > > > > (try_run):
>> > > > > > >   Policy CMP0067 is not set: Honor language standard in
>> > > > > try_compile()
>> > > > > > >   source-file signature.  Run "cmake --help-policy  
>> CMP0067"
>> > > for
>> > > > > policy
>> > > > > > >   details.  Use the cmake_policy command to set the  
>> policy and
>> > > > > > > suppress this
>> > > > > > >   warning.
>> > > > > > >
>> > > > > > >   For compatibility with older versions of CMake,  
>> try_compile
>> > > is
>> > > > > not
>> > > > > > > honoring
>> > > > > > >   language standard variables in the test project:
>> > > > > > >
>> > > > > > >     CMAKE_CXX_STANDARD
>> > > > > > >     CMAKE_CXX_STANDARD_REQUIRED
>> > > > > > >     CMAKE_CXX_EXTENSIONS
>> > > > > > >
>> > > > > > > Call Stack (most recent call first):
>> > > > > > >   CMakeLists.txt:242 (find_package)
>> > > > > > > This warning is for project developers.  Use -Wno-dev to
>> > > suppress
>> > > > > it.
>> > > > > > >
>> > > > > > >    Called from: [2]
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/cmake/modules/FindLibical.cmake
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > CMake Error at cmake/modules/FindLibical.cmake:75  
>> (message):
>> > > > > > >   Unable to compile or run the libical version detection
>> > > program.
>> > > > > > > Call Stack (most recent call first):
>> > > > > > >   CMakeLists.txt:242 (find_package)
>> > > > > > >
>> > > > > > >
>> > > > > > >    Called from: [2]
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/cmake/modules/FindLibical.cmake
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > -- Configuring incomplete, errors occurred!
>> > > > > > > See also
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> "/usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/build/CMakeFiles/CMakeOutput.log".
>> > > > > > > See also
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> "/usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/build/CMakeFiles/CMakeError.log".
>> > > > > > > make: *** No targets specified and no makefile found.   
>> Stop.
>> > > > > > >
>> > > > > > >  
>> =============================================================
>> > > > > > >
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > -- Found PythonLibs: //lib64/libpython2.7.so (found  
>> suitable
>> > > > > version
>> > > > > > > "2.7.15", minimum required is "2.7.15")
>> > > > > > >    Called from: [4]
>> > > > > /usr/share/cmake/Modules/FindPackageMessage.cmake
>> > > > > > >                 [3]
>> > > > > > >  
>> /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake
>> > > > > > >                 [2]
>> > > /usr/share/cmake/Modules/FindPythonLibs.cmake
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > -- Could NOT find PY_weboob (missing: PY_WEBOOB)
>> > > > > > >    Called from: [4]
>> > > > > > >  
>> /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake
>> > > > > > >                 [3]
>> > > > > > >  
>> /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake
>> > > > > > >                 [2]
>> > > > > > >
>> > > > > >
>> > > > >
>> > > >
>> > >
>> >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/cmake/modules/FindPythonModule.cmake
>> > > > > > >                 [1]
>> > > > > > >
>> > >  
>> /usr/local/src/kmm/kmymoney-5.0-2018.11.07-18:23:12/CMakeLists.txt
>> > > > > > > -- Configuring incomplete, errors occurred!
>> > > > > > >
>> > > > > > >
>> > > > > > > ----
>> > > > > > > Brendan Coupe
>> > > > > > > On Wed, Nov 7, 2018 at 5:42 PM Jack
>> > > > > <ostroffjh at users.sourceforge.net>
>> > > > > > > wrote:
>> > > > > > > >
>> > > > > > > > On 2018.11.07 19:22, Brendan Coupe wrote:
>> > > > > > > > > weboob errors seem to be unfixable in Fedora. I  
>> think they
>> > > > > have
>> > > > > > > been
>> > > > > > > > > around for a while, I found them in previous  
>> requests for
>> > > > > help. I
>> > > > > > > > > don' think they have anything to do with the compile
>> > > failing.
>> > > > > not
>> > > > > > > > > sure if they cause any of the weird behavior I was
>> > > seeing. I
>> > > > > will
>> > > > > > > try
>> > > > > > > > > to resurrect an old F28 system. I'm hoping the repos  
>> have
>> > > > > 5.0.2
>> > > > > > > soon
>> > > > > > > > > since I'm currently running 5.0.1 and dealing with  
>> old
>> > > > > problems
>> > > > > > > that
>> > > > > > > > > have been fixed.
>> > > > > > > > I agree weboob is not relevant here.
>> > > > > > > > >
>> > > > > > > > > Is this the detection program you are talking about:
>> > > > > > > > > cmake/modules/FindLibical.cmake
>> > > > > > > > No, the .cmake program tells cmake where to look for  
>> stuff,
>> > > and
>> > > > > how
>> > > > > > > to
>> > > > > > > > set various variables used later in the cmake run.  To
>> > > actually
>> > > > > > > check,
>> > > > > > > > cmake creates a .c program (using hints and info from  
>> the
>> > > .cmake
>> > > > > > > file,
>> > > > > > > > and then tries to compile and link it.  (Some checks  
>> might
>> > > be
>> > > > > done
>> > > > > > > with
>> > > > > > > > pkg_config)  Normally that fails if libical isn't  
>> found, or
>> > > > > isn't a
>> > > > > > > > high enough version, for example.  Your error says "  
>> Unable
>> > > to
>> > > > > > > compile
>> > > > > > > > or run the libical version detection program." which  
>> seems
>> > > like
>> > > > > a
>> > > > > > > > different sort of problem.  I would look through the  
>> cmake
>> > > > > --help
>> > > > > > > > output to see what parameters might provide additional
>> > > details
>> > > > > about
>> > > > > > > > what is actually failing.
>> > > > > > > > >
>> > > > > > > > > Line 75 is the source of the message.
>> > > > > > > > > gcc.x86_64                                   
>> 8.2.1-4.fc29
>> > > > > > > > >          @updates
>> > > > > > > > > gcc-c++.x86_64                               
>> 8.2.1-4.fc29
>> > > > > > > > >          @updates
>> > > > > > > > > gcc-gdb-plugin.x86_64                        
>> 8.2.1-4.fc29
>> > > > > > > > >          @updates
>> > > > > > > > > libgcc.x86_64                                
>> 8.2.1-4.fc29
>> > > > > > > > >          @updates
>> > > > > > > > This, however, might be the issue.  I'll have to try
>> > > installing
>> > > > > gcc
>> > > > > > > 8
>> > > > > > > > and see how it works for me.  As a temporary fix, you  
>> might
>> > > try
>> > > > > > > using
>> > > > > > > > gcc 7.
>> > > > > > > >
>> > > > > > > > Jack
>> > > > > > > > >
>> > > > > > > > > ----
>> > > > > > > > > Brendan Coupe
>> > > > > > > > > On Wed, Nov 7, 2018 at 4:47 PM Jack
>> > > > > > > <ostroffjh at users.sourceforge.net>
>> > > > > > > > > wrote:
>> > > > > > > > > >
>> > > > > > > > > > Hello Brendan,
>> > > > > > > > > >
>> > > > > > > > > > On 2018.11.07 18:17, Brendan Coupe wrote:
>> > > > > > > > > > > I have been compiling from source for 10+ years.  
>> I had
>> > > > > been
>> > > > > > > able
>> > > > > > > > > to
>> > > > > > > > > > > compile both the master branch and the 5.0  
>> branch on
>> > > my
>> > > > > > > Fedora 28
>> > > > > > > > > > > system until a month or so ago. I was having some
>> > > weird
>> > > > > > > behavior
>> > > > > > > > > in
>> > > > > > > > > > > KMM5 that was not affecting a clean Fedora 28  
>> test VM
>> > > > > running
>> > > > > > > KMM
>> > > > > > > > > > > from the repos so I decided to build a new clean
>> > > Fedora 29
>> > > > > > > system.
>> > > > > > > > > > It may be worth checking the release notes for  
>> Fedora
>> > > 29 to
>> > > > > see
>> > > > > > > what
>> > > > > > > > > > changed, and if anything looks possibly relevant.
>> > > > > > > > > >
>> > > > > > > > > > > I also upgraded the old system from Fedora 28 to  
>> 29.
>> > > Both
>> > > > > > > systems
>> > > > > > > > > > > fail at the same point while compiling from the  
>> master
>> > > > > and 5.0
>> > > > > > > > > > > branches:
>> > > > > > > > > > It certainly sounds like some difference between  
>> 28 and
>> > > 29
>> > > > > is
>> > > > > > > behind
>> > > > > > > > > > the failure.
>> > > > > > > > > >
>> > > > > > > > > > > -- Found LibOfx:
>> > > > > > > > > > > -- Performing Test LIBOFX_HAVE_CLIENTUID
>> > > > > > > > > > > -- Performing Test LIBOFX_HAVE_CLIENTUID -  
>> Success
>> > > > > > > > > > > -- Found PythonInterp: /usr/bin/python2 (found
>> > > suitable
>> > > > > > > version
>> > > > > > > > > > > "2.7.15", minimum required is "2.6")
>> > > > > > > > > > > -- Found PythonLibs: /usr/lib64/libpython2.7.so  
>> (found
>> > > > > > > suitable
>> > > > > > > > > > > version "2.7.15", minimum required is "2.7.15")
>> > > > > > > > > > > -- Could NOT find PY_weboob (missing: PY_WEBOOB)
>> > > > > > > > > > I'm assuming this is not an issue for you.
>> > > > > > > > > > > CMake Error at cmake/modules/FindLibical.cmake:75
>> > > > > (message):
>> > > > > > > > > > >   Unable to compile or run the libical version
>> > > detection
>> > > > > > > program.
>> > > > > > > > > > This does seem odd.  It appears not even getting  
>> to the
>> > > > > point of
>> > > > > > > > > > actually checking for an installed libical.
>> > > > > > > > > > > Call Stack (most recent call first):
>> > > > > > > > > > >   CMakeLists.txt:242 (find_package)
>> > > > > > > > > > At the point of failure, check in the build  
>> directory.
>> > > > > Often
>> > > > > > > you
>> > > > > > > > > can
>> > > > > > > > > > find the libical version detection program (I've
>> > > forgotten
>> > > > > > > whether
>> > > > > > > > > it
>> > > > > > > > > > leaves the .c program around, and see if  
>> configure.log
>> > > (or
>> > > > > > > something
>> > > > > > > > > > like it) might have a more detailed reason for the
>> > > failure
>> > > > > to
>> > > > > > > > > compile
>> > > > > > > > > > that .c program.
>> > > > > > > > > >
>> > > > > > > > > > Not sure if it might be relevant, but what version  
>> of
>> > > gcc
>> > > > > are
>> > > > > > > you
>> > > > > > > > > > using, both in F28 and in F29?
>> > > > > > > > > > >
>> > > > > > > > > > > -- Configuring incomplete, errors occurred!
>> > > > > > > > > > > ========================================
>> > > > > > > > > > >
>> > > > > > > > > > > libical appears to be installed:
>> > > > > > > > > > >
>> > > > > > > > > > >
>> > > > > > > > > > > dnf list installed | grep libical
>> > > > > > > > > > > libical.x86_64
>> > > > > > > > > 3.0.3-7.fc29
>> > > > > > > > > > >                                  @fedora
>> > > > > > > > > > > libical-devel.x86_64
>> > > > > > > > > 3.0.3-7.fc29
>> > > > > > > > > > >                                  @fedora
>> > > > > > > > > > >
>> > > > > > > > > > > Any ideas?
>> > > > > > > > > > >
>> > > > > > > > > > > Thanks,
>> > > > > > > > > > >
>> > > > > > > > > > > ----
>> > > > > > > > > > > Brendan Coupe
>> > > > > > > > > > Jack
>> > >
>> >
>> >
>> 
>