Re: [Plplot-general] mingw32-make[2]: *** No rule to make target `drivers/CMakeFiles/null.dir/progres s.make'. Stop.

[Alan W. I. <ir...@be...>]

Hi Frank:

I will let those familiar with wxwidgets on Windows comment on the
linking errors you encountered in that case, but I suggest that
you also try building a simple PLplot with minimal external
dependencies to make sure the core library is working OK.
I give a cookbook for such a simple build and test below.

On 2011-10-03 15:07-0000 FrankTheFox wrote:

> ENABLE_DYNDRIVERS:      ON
> DRIVERS_LIST: mem;null;ps;svg;wingcc;wxwidgets;xfig
> DEVICES_LIST: mem;null;ps;svg;wingcc;wxwidgets;xfig

For the simple test case, eliminate wingcc and wxwidgets by using the
cmake options -DPLD_wingcc=OFF -DPLD_wxwidgets=OFF Also, for the
simple case I would eliminate all bindings other than C and C++ using
the combination of cmake options -DDEFAULT_NO_BINDINGS=ON
-DENABLE_cxx=ON

> And CMakeError.log file reports:

That file always shows lots of uninteresting errors because what CMake
does for system introspection is a lot of different tests.  If
something fails, it notes it often finds other alternatives, and
adjusts the build accordingly.  So I would only pay attention to that
file in case of some error that was impossible to understand
otherwise.

In any case, please attach files separately rather than mixing them in
with your e-mail.  That allows people to read your e-mail without
necessarily looking at the detailed results. Also, it was hard to tell
where the CMakeError.log file quit, and the results of your actual
build started.  I assume that the build output (as opposed to cmake
output and CMakeError.log output) begins where you said the following
in your e-mail

> In my console window i see:
>
>
> Linking CXX shared module ..\dll\wxwidgets.dll

[...]

If so, it means you only captured a small part of the build output. 
What you need to do is capture it all.  On bash (which is available
with MSYS) all you would need to do to capture all output is

cmake >& cmake.out
make >& make.out

etc.

I assume there are similar ways of doing that reliably in a
non-bash environment, but I don't know what they are.

Note, that the "MinGW Makefiles" generator for cmake will not work if
"sh.exe" which is an alias for "bash.exe" is on your PATH.  On the
other hand, if you use the "MSYS Makefiles" generator, then sh.exe and
bash.exe must be on your PATH.  Note that generator is a more natural fit to
the PLplot build since it also allows you to test PLplot (which is done
with several bash scripts that are run automatically as part of the
testing procedures, see below).

Getting back to the simple test case I was talking about before, here is
the cookbook that I suggest.

1. Download the latest MinGW/MSYS, and put the MinGW part of it on
your PATH to get access to those compilers, and the MSYS part of it on
your PATH so that MSYS make and MSYS bash are both accessible.

2. Configure a simple PLplot using the "MSYS Makefiles" generator
including test infrastructure (-DBUILD_TEST=ON) for that case.

cmake -G "MSYS Makefiles" \
-DPLD_wingcc=OFF -DPLD_wxwidgets=OFF \
-DDEFAULT_NO_BINDINGS=ON -DENABLE_cxx=ON \
-DBUILD_TEST=ON /path/to/top/directory/of/source/tree >& cmake.out

3. Put the absolute path of the dll subdirectory of the top-level
build-tree directory on your PATH.  That's where all the dll's will
reside as a result of the build part of the next step.

4. Build and test the result

make test_noninteractive >& make_test_interactive.out

Build and test are combined together by this target because
of the nice dependencies created for make by the cmake configuration
step.

Look for any errors in those *.out files. Also, please send the
cmake.out and make_test_interactive.out files to me as attachments to
a private e-mail so I can confirm whatever you find, compare with
my wine results, etc.

If that simple build and test works well, then that gives lots of confidence
that the only remaining issue for you is the wxwidgets linking issue.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state
implementation for stellar interiors (freeeos.sf.net); the Time
Ephemerides project (timeephem.sf.net); PLplot scientific plotting
software package (plplot.sf.net); the libLASi project
(unifont.org/lasi); the Loads of Linux Links project (loll.sf.net);
and the Linux Brochure Project (lbproject.sf.net).
__________________________

Linux-powered Science
__________________________