[Kst] kst 1.4.0 compilation on Solaris

[Brisset, Nicolas]

Hi,

I have compiled kst 1.4.0 on a Solaris 8 machine (with a hand-installed
KDE 3.4.0 and gcc 3) and I had to tweak a couple of things. It would be
nice if we could fix them properly, but at least I'm posting the
problems to the list for the record. It may even help one or two
adventurous people out there who might be running kst on Solaris :-)
The problems were the following:
1) I had to set SHELL = /bin/bash manually in configure (around line
245) otherwise libtool failed without any explicit error message. The
/bin/sh shipped with Solaris apparently doesn't handle all tests
correctly
2) I had to remove a #ifdef NAN from kstmath.h as apparently Solaris
defines NAN, but not all that is required. Below is the corresponding
diff:
--- kst-1.4.0/kst/src/libkst/kstmath.h  2007-03-15 22:20:54.000000000
+0100
+++ kst-1.4.0/kst/src/libkst/kstmath.h.mod      2007-06-11
14:39:28.716519000 +0200
@@ -35,7 +35,6 @@
 ** For systems without NAN, this is a NAN in IEEE double format.
 ** Code from KDE's kjs libarary.
 */
-#if !defined(NAN)
 #include <qconfig.h>
 static double nan__()
 {
@@ -54,7 +53,6 @@
   return NaN;
 }
 #define NAN (::nan__())
-#endif

 /*
 ** Both Solaris and FreeBSD-current do weird things with the
3) I wanted to add a custom datasource, which I used to compile out of
the source tree. As the out-of-source-tree datasource template no longer
works, it means one has to add the datasource to the autotools-managed
build process, which is not really straightforward for newbies :-( I was
thinking that I should maybe add a text doc somewhere to document the
steps in case someone needs it. What do you think ? The other option
would be to upgrade the stand-alone datasource template, but with all
the changes coming I don't know whether it's worth the effort.

Nicolas