Re: [Plplot-general] Problem building 5.11 on Windows 8.1

[Walt B. <wal...@gm...>]

Here are my attempts to run the test as you requested.

Just to get all the facts:

The TDM version of Mingw is in C:\mingw.
It includes gcc version 5.1.

Msys is in C:\msys\1.0--the MSYS web site implied this
was not the latest, but I couldn't find anything else.

I enclose the test output with the directory "shared" removed.
OW my mail wouldn't let me attach the tgz result.

How do I install libplplot.dll if I can't build plplot?

There may simply be too many things different, especially gcc 4.9 vs 5.1,
but I will keep trying things if it will help you improve Plplot.

BTW, I tried a build with MSYS instead of MINGW Makefiles for the first
time before running the test and got pretty much the same results as before.

[image: Inline image 1]



On Mon, Jul 13, 2015 at 11:15 AM, Arjen Markus <Arj...@de...>
wrote:

>  Hi Walt,
>
>
>
> As Alan suggested off-list, it might be a good idea to run the
> comprehensive_test.sh script yourself and send me and Alan the resulting
> tarball. Here is the script I use to get it going:
>
>
> export PATH=/d/cmake/bin:$PATH
>
>
>
> # Generate all script results in ../comprehensive_test_disposeable
>
> ../plplot-git/scripts/comprehensive_test.sh
> --do_test_traditional_install_tree no --do_test_interactive no --do_ctest
> no --generator_string "MSYS Makefiles" --build_command make --ctest_command
> ctest --do_clean_as_you_go no --cmake_added_options "-DENABLE_tcl=OFF"
> --do_shared yes --do_nondynamic yes
>
>
>
> The command line is longer than it actually needs to be (it has gathered a
> lot of cruft over the various iterations), but I have it stored in a small
> shell script for easy use.
>
>
>
> The output is gathered in a concise tarball in the
> comprehensive_test_disposeable directory. It is called
>
> “comprehensive_test.tar.gz”.
>
>
>
> The nice thing about this script, besides being an automatic test of
> various build options (you only need to confirm that you want to run it,
> because it will throw away anything in the test directory), is that it
> collects as much information as could be relevant from the environment.
> That will save us some time ;).
>
>
>
> Regards,
>
>
>
> Arjen
>
>
>
>
>
>
>
> *From:* Arjen Markus
> *Sent:* Monday, July 13, 2015 1:18 PM
> *To:* Arjen Markus; Walt Brainerd
> *Cc:* plplot_general
> *Subject:* RE: [Plplot-general] Problem building 5.11 on Windows 8.1
>
>
>
> Hi Walt,
>
>
>
> The link problem with Fortran under MinGW I reported is solved – it turned
> out that the condition to include the .def file in the link step was
> incomplete. With that fixed it works all fine now. For this I have tested
> the build options covered by the comprehensive test script, using the GCC
> compiler suite that comes with MSYS2.
>
>
>
> Now, since you have problems with theTDM version of the GCC suite, can you
> tell me how the make process fails? What are the error messages? It seems
> strange that it would succeed after several retries – it would almost seem
> that some dependencies are not handled correctly. Or perhaps there is a
> problem with parallel builds? I have done my tests with a single-threaded
> make command.
>
>
>
> Regards,
>
>
>
> Arjen
>
>
>
>
>
> *From:* Arjen Markus [mailto:Arj...@de...
> <Arj...@de...>]
> *Sent:* Monday, July 13, 2015 10:35 AM
> *To:* Walt Brainerd
> *Cc:* plplot_general
> *Subject:* Re: [Plplot-general] Problem building 5.11 on Windows 8.1
>
>
>
> Hi Walt,
>
>
>
> I had to reinstall MinGW 64-bits, as indeed it was hardly useable. I have
> done so by installing MSYS2 as recommended by Alan. Now I see problems with
> the .def file for the Fortran libraries – it has to do with the calling
> convention (stdcall versus cdecl and that sort of fun stuff). Is that the
> sort of failures you have been seeing? Here is a small fragment:
>
> Cannot export _PLPARSEOPTS@4: symbol not defined
>
> Cannot export _PLPLOTP_mp_PLAXES@40: symbol not defined
>
> Cannot export _PLPLOTP_mp_PLBOX3@72: symbol not defined
>
> Cannot export _PLPLOTP_mp_PLBOX@32: symbol not defined
>
> Cannot export _PLPLOTP_mp_PLCOLORBAR_1@96: symbol not defined
>
>
>
> It is not all that difficult to repair, but it has to be done, otherwise
> the shared library option won’t work.
>
>
>
> Regards,
>
>
>
> Arjen
>
>
>
> *From:* Walt Brainerd [mailto:wal...@gm...
> <wal...@gm...>]
> *Sent:* Friday, July 10, 2015 9:27 PM
> *To:* Arjen Markus
> *Cc:* plplot_general
> *Subject:* Re: [Plplot-general] Problem building 5.11 on Windows 8.1
>
>
>
> Alan: mingw-64 etc. does not yet include gcc 5.1;
>
> that is why I am using the TDM version.
>
>
>
> Arjen: with the new version, plplot builds (the static
>
> library version), but when I run it, I get all kinds of
>
> errors about missing routines of the form __imp*.
>
> being called from wingcc (for example).
>
>
>
> So I tried the DLL version (which is what I think I
>
> was successful with before). make was strange.
>
> I would get an error, then try it again. It got further
>
> before getting another error. After about 5-6 tries
>
> it went all the way.
>
>
>
> Trying to compile a code then produced an error
>
> trying to load wingcc.dll. Windows says it isn't
>
> suitable to run on Windows; in a bash shell, it
>
> just says there is an error loading wingcc.dll.
>
>
>
> Any other suggestions?
>
>
>
> On Fri, Jul 10, 2015 at 2:14 AM, Arjen Markus <Arj...@de...>
> wrote:
>
> Hi Walt,
>
>
>
> I just completed the set of comprehensive tests and committed the small
> patch I had to make for the access() function. With this latest version it
> ought to work for you too.
>
>
> Regards,
>
>
>
> Arjen
>
>
>
>
>
> *From:* Arjen Markus [mailto:Arj...@de...]
> *Sent:* Friday, July 10, 2015 10:37 AM
> *To:* Walt Brainerd; plplot_general
> *Subject:* Re: [Plplot-general] Problem building 5.11 on Windows 8.1
>
>
>
> Hi Walt,
>
>
>
> I am trying this myself – using the comprehensive tests script under MSYS
> – and I ran into the same issue. It seems that the logic for including the
> header file unistd.h is incomplete or incorrect as far as MinGW is
> concerned. I have changed this in my own repository to:
>
>
>
>
>
> #include "plConfig.h"
>
> #if !PL_HAVE_UNISTD_H
>
>   #define F_OK    1
>
>   #include <stdio.h>
>
> int access( char *filename, int flag )
>
>
>
> and that is doing the trick as far as the build up to now is concerned. If
> successful, I will contribute this.
>
>
>
> Regards,
>
>
>
> Arjen
>
>
>
>
>
> *From:* Walt Brainerd [mailto:wal...@gm...]
> *Sent:* Friday, July 10, 2015 1:25 AM
> *To:* plplot_general
> *Subject:* [Plplot-general] Problem building 5.11 on Windows 8.1
>
>
>
> I have the TDM version of MinGW installed.
>
>
>
> C:\Users\Walt>gfortran --version
>
> GNU Fortran (tdm64-1) 5.1.0
>
> Copyright (C) 2015 Free Software Foundation, Inc.
>
>
>
> I downloaded 5.11.0 using git and built with the following
>
> (all one line, of course):
>
>
>
> cmake -G "MinGW Makefiles" -DCMAKE_INSTALL_PREFIX=install
>
> -DBUILD_SHARED_LIBS=OFF -DDEFAULT_NO_BINDINGS=ON -DENABLE_cxx=ON
>
> -DENABLE_f95=ON -DDEFAULT_ALL_DEVICES=ON ../plplot.git
>
>
>
> [Something different: I got a lot more devices and drivers.]
>
>
>
> Make produces the following:
>
>
>
> C:\walt\Software\Plplot\BUILD>make
>
> [  0%] Built target csirocsa
>
> [  3%] Built target deltaT-gen
>
> [  4%] Built target deltaT.h_built
>
> [  6%] Built target tai-utc-gen
>
> [  7%] Built target tai-utc.h_built
>
> [ 10%] Built target qsastime
>
> [ 12%] Built target nistcd
>
> [ 12%] Built target cdexpert
>
> [ 13%] Built target cdmulti
>
> [ 15%] Built target cdsimple
>
> [ 16%] Built target cdtest
>
> [ 18%] Built target cdtext
>
> [ 18%] Built target color16
>
> [ 26%] Built target test_nistcd
>
> [ 26%] Built target plhershey-unicode-gen
>
> [ 27%] Built target plhershey-unicode.h_built
>
> [ 29%] Building C object src/CMakeFiles/plplot.dir/plfreetype.c.obj
>
> In file included from
> C:/Fortran_Tools/gfortran/x86_64-w64-mingw32/include/unistd.h:10:0,
>
>                  from
> C:/walt/Software/Plplot/plplot.git/include/plplotP.h:110,
>
>                  from
> C:\walt\Software\Plplot\plplot.git\src\plfreetype.c:90:
>
> C:/Fortran_Tools/gfortran/x86_64-w64-mingw32/include/io.h:310:15: error:
> conflicting types for 'access'
>
>    int __cdecl access(const char *_Filename,int _AccessMode)
> __MINGW_ATTRIB_DEPRECATED_MSVC2005;
>
>                ^
>
> C:\walt\Software\Plplot\plplot.git\src\plfreetype.c:69:5: note: previous
> definition of 'access' was here
>
>  int access( char *filename, int flag )
>
>      ^
>
> In file included from
> C:\walt\Software\Plplot\plplot.git\src\plfreetype.c:90:0:
>
> C:/walt/Software/Plplot/plplot.git/include/plplotP.h:252:0: warning:
> "isfinite" redefined
>
>  #  define isfinite    finite
>
>  ^
>
> In file included from
> C:/walt/Software/Plplot/plplot.git/include/plplotP.h:105:0,
>
>                  from
> C:\walt\Software\Plplot\plplot.git\src\plfreetype.c:90:
>
> C:/Fortran_Tools/gfortran/x86_64-w64-mingw32/include/math.h:520:0: note:
> this is the location of the previous definition
>
>  #define isfinite(x) ((fpclassify(x) & FP_NAN) == 0)
>
>  ^
>
> src\CMakeFiles\plplot.dir\build.make:366: recipe for target
> 'src/CMakeFiles/plplot.dir/plfreetype.c.obj' failed
>
> make[2]: *** [src/CMakeFiles/plplot.dir/plfreetype.c.obj] Error 1
>
> CMakeFiles\Makefile2:798: recipe for target
> 'src/CMakeFiles/plplot.dir/all' failed
>
> make[1]: *** [src/CMakeFiles/plplot.dir/all] Error 2
>
> Makefile:135: recipe for target 'all' failed
>
> make: *** [all] Error 2
>
>
>
> To me, the first one (access) looks like a conflict between gcc and plplot.
>
> The second (isfinite) is a conflict between two things in Plplot???
>
>
>
> Help please?  Thanks.
>
>
>
>
>
> --
>
> Walt Brainerd
>
> DISCLAIMER: This message is intended exclusively for the addressee(s) and
> may contain confidential and privileged information. If you are not the
> intended recipient please notify the sender immediately and destroy this
> message. Unauthorized use, disclosure or copying of this message is
> strictly prohibited. The foundation 'Stichting Deltares', which has its
> seat at Delft, The Netherlands, Commercial Registration Number 41146461, is
> not liable in any way whatsoever for consequences and/or damages resulting
> from the improper, incomplete and untimely dispatch, receipt and/or content
> of this e-mail.
>
> DISCLAIMER: This message is intended exclusively for the addressee(s) and
> may contain confidential and privileged information. If you are not the
> intended recipient please notify the sender immediately and destroy this
> message. Unauthorized use, disclosure or copying of this message is
> strictly prohibited. The foundation 'Stichting Deltares', which has its
> seat at Delft, The Netherlands, Commercial Registration Number 41146461, is
> not liable in any way whatsoever for consequences and/or damages resulting
> from the improper, incomplete and untimely dispatch, receipt and/or content
> of this e-mail.
>
>
>
>
>
> --
>
> Walt Brainerd
>
> DISCLAIMER: This message is intended exclusively for the addressee(s) and
> may contain confidential and privileged information. If you are not the
> intended recipient please notify the sender immediately and destroy this
> message. Unauthorized use, disclosure or copying of this message is
> strictly prohibited. The foundation 'Stichting Deltares', which has its
> seat at Delft, The Netherlands, Commercial Registration Number 41146461, is
> not liable in any way whatsoever for consequences and/or damages resulting
> from the improper, incomplete and untimely dispatch, receipt and/or content
> of this e-mail.
>   DISCLAIMER: This message is intended exclusively for the addressee(s)
> and may contain confidential and privileged information. If you are not the
> intended recipient please notify the sender immediately and destroy this
> message. Unauthorized use, disclosure or copying of this message is
> strictly prohibited. The foundation 'Stichting Deltares', which has its
> seat at Delft, The Netherlands, Commercial Registration Number 41146461, is
> not liable in any way whatsoever for consequences and/or damages resulting
> from the improper, incomplete and untimely dispatch, receipt and/or content
> of this e-mail.
>



-- 
Walt Brainerd