Hello,

while testing the patch Jim Dishaw provided I ran into
a few oddities that could actually be considered nasty:

1. The file "plplotf95.def", the list of subroutines and
   functions that must be exported is NOT seen as a
   dependency. So only via "make clean" will a change here
   force a rebuild of the libraries.

2. The executables are linked with a flag /FORCE (I am
   talking of the combination MS Visual C/C++ and Compaq
   Visual Fortran under Windows XP here). This means that
   even if there are unresolved externals (subroutines
   not exported from the library for instance) a program
   is created. This may run correctly up to the point
   that one of the missing routines/functions is
   called.
   The alternative (not having /FORCE) may present even
   more distasteful problems: the linker will bail out
   on duplicate entries from default libraries ...


3. If g95 or gfortran is installed and reachable via the
   path, then these compilers will be taken instead of
   CVF (or Intel Fortran, I guess), but they do not
   play well with MSVC.
   I posted this problem on the CMake mailing list a
   few weeks ago and it turns out to be a rather nasty
   problem. The people developing CMake are looking
   into it, but the issue may have far-reaching
   consequences:
   Essentially, the various languages
   are treated separately now and the various CMake
   variables are language-neutral. To solve this
   issue the simplest but incompatible change is
   to make the variables language-specific.

(This mail has become a bit longer than I anticipated,
but I hope the issues are clear.)

Regards,

Arjen

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Plplot-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/plplot-devel

Reply via email to