[Plplot-cvs] SF.net SVN: plplot:[13064] trunk/cmake/modules/double.cmake

[<ai...@us...>]

Revision: 13064
          http://sourceforge.net/p/plplot/code/13064
Author:   airwin
Date:     2014-03-16 00:35:36 +0000 (Sun, 16 Mar 2014)
Log Message:
-----------
Force LIB_TAG to be the empty string as discussed on list.

Tested on Linux using the test_interactive and test_noninteractive
targets in the build tree.

Modified Paths:
--------------
    trunk/cmake/modules/double.cmake

Modified: trunk/cmake/modules/double.cmake
===================================================================
--- trunk/cmake/modules/double.cmake	2014-03-16 00:26:59 UTC (rev 13063)
+++ trunk/cmake/modules/double.cmake	2014-03-16 00:35:36 UTC (rev 13064)
@@ -25,7 +25,17 @@
 option(PL_DOUBLE 
 "Compile the library with double precision floating point" ON)
 
-if(PL_DOUBLE)
-  set(LIB_TAG "${LIB_TAG}d")
-endif(PL_DOUBLE)
+# Force LIB_TAG to be empty in all cases.  This is completely
+# different from the previous case where "d" was appended to the user
+# setting of this variable if the library was double precision.  As
+# discussed on list we are forcing LIB_TAG to be empty because it is
+# only applied to a subset of our libraries (because the other ones do
+# not have a single-precision or double-precision variant).
+# Furthermore, the recommended and absolutely most reliable way to
+# distinguish between various kinds of PLplot builds is not by
+# appending various confusing library suffixes.  (For example, "d" is
+# often incorrectly assumed to refer to a debugging version of the
+# PLplot library.)  Instead, simply use a different install prefix for
+# each different kind of build.
+set(LIB_TAG "" CACHE STRING "LIB_TAG should _always_ be the empty string" FORCE)
 

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.