On 2009-02-26 21:20+0100 Werner Smekal wrote:
> Hi Alan and Andrew,
>
> in plplot.cmake we determine if finite/isnan/isinf or
> _finite/_isnan/_isinf is available with such code
>
> check_symbol_exists(finite "math.h" HAVE_FINITE)
> if(NOT HAVE_FINITE)
> check_function_exists(finite HAVE_FINITE2)
> endif(NOT HAVE_FINITE)
> set(HAVE_FINITE ${HAVE_FINITE2})
> if(NOT HAVE_FINITE)
> check_symbol_exists(_finite "math.h" _HAVE_FINITE)
> if(NOT _HAVE_FINITE)
> check_function_exists(_finite _HAVE_FINITE2)
> endif(NOT _HAVE_FINITE)
> set(_HAVE_FINITE ${_HAVE_FINITE2})
> set(HAVE_FINITE ${_HAVE_FINITE} CACHE INTERNAL "Have function _finite")
> endif(NOT HAVE_FINITE)
There are a number of problems in the above logic.
(*) HAVE_FINITE, HAVE_FINITE2, _HAVE_FINITE and _HAVE_FINITE2 are all cached
as a result of check_symbol_exists and check_function_exists. (See the
commands
SET(${VARIABLE} 1 CACHE INTERNAL "Have symbol ${SYMBOL}")
and
SET(${VARIABLE} 1 CACHE INTERNAL "Have function ${FUNCTION}")
in CheckSymbolExists.cmake and CheckFunctionExists.cmake). So all further
attempts to set those variables should also mention CACHE INTERNAL or else
you should use a different name (UNCACHED_HAVE_FINITE ?) for the uncached
variables you are setting. (In CMake the same name can refer to separate
variables, the cached one and the uncached one.) It's possible the current
logic works, but with this reuse of names it gets quite confusing so I would
advocate using separate names for variables that mean separate things, and
definitely you should not use the same name for both a cached variable and
uncached variable.
(*) The
set(HAVE_FINITE ${HAVE_FINITE2})
assumes check_function_exists(finite HAVE_FINITE2) has been called, but
what if the first check_symbol_exists succeeds? Then, that success is
ignored, HAVE_FINITE2 is undefined, and then you start looking incorrectly
for the _finite symbol or _finite function.
Once those two issues are straightened out, I think the logic will work both
the first manual time and subsequent automatic times cmake is called. Note,
on the second and subsequent times, the cache is used to retrieve values
rather than actually doing the checks to determine those values. That is
why CMake is so notorious for speed, but it also means you have to be
careful of whether variables are cached or not like I advocate above.
> Anyway, do you both can help me to solve this problem, since I'm not
> sure how to proceed.
I would suggest the above changes. Then debug the second and subsequent
times cmake is run by putting in lots of message statements to verify all
values are being set correctly from the cache.
Good luck sorting this out.
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); PLplot scientific plotting software
package (plplot.org); 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
__________________________
------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
Plplot-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/plplot-devel
