ctype checks not falling out soon enough (peudopatch)
Phil Edwards
pedwards@jaj.com
Sat Apr 1 00:00:00 GMT 2000
The ctype checks are correctly determining that I'm on a Solaris 7 box.
Unfortunately they then keep going and determine that I'm on an AIX box.
:-) I've a patch that only checks for <foo>'s ctype if the default would
otherwise still be used.
The checks got indented, which means the diff looks hideous. Here is the
diff with -b, to show the essential changes. If y'all think this is good,
then I'll check in the real patch, along with the regenerated files.
Phil
(It still doesn't build due to signed/unsigned comparisons in ctype.cc,
but that's a different problem.)
diff -ru3b libstdc++/acinclude.m4 l3/acinclude.m4
--- libstdc++/acinclude.m4 Tue Mar 14 14:58:34 2000
+++ l3/acinclude.m4 Tue Mar 14 17:07:57 2000
@@ -243,6 +243,7 @@
dnl If doesn't match any specified, go with defaults.
ctype_default=yes
+ if test $ctype_default = yes; then
dnl Test for <ctype> functionality -- gnu-linux
AC_MSG_CHECKING([for gnu-linux <ctype>])
AC_TRY_COMPILE([#include <ctype.h>],
@@ -257,7 +258,9 @@
ctype_include_dir="config/gnu-linux"
ctype_default=no
fi
+ fi
+ if test $ctype_default = yes; then
dnl Test for <ctype> functionality -- solaris 2.5.1
AC_MSG_CHECKING([for solaris 2.5.1 <ctype>])
AC_TRY_COMPILE([#include <ctype.h>],
@@ -300,7 +303,9 @@
ctype_default=no
fi
fi
+ fi
+ if test $ctype_default = yes; then
dnl Test for <ctype> functionality -- aix
AC_MSG_CHECKING([for aix <ctype>])
AC_TRY_COMPILE([#include <ctype.h>],
@@ -315,7 +320,9 @@
ctype_include_dir="config/aix"
ctype_default=no
fi
+ fi
+ if test $ctype_default = yes; then
dnl Test for <ctype> functionality -- newlib
AC_MSG_CHECKING([for newlib <ctype>])
AC_TRY_COMPILE([#include <ctype.h>],
@@ -329,6 +336,7 @@
ctype_include_dir="config/newlib"
ctype_default=no
fi
+ fi
if test $ctype_default = "yes"; then
ctype_include_dir="config/default"
More information about the Libstdc++
mailing list