Hi all.
I don't follow why ${host_alias} is being used in the toplevel
apr configure.in file when the ${OS} variable is used in other
places. The OS variable is set to ${host} which is expanded
from the value passed in on the configure command line after
being filtered through config.sub.
For example:
% ./config.sub mingw32
i386-pc-mingw32
% ./configure --host=mingw32
loading cache ./config.cache
checking host system type... i386-pc-mingw32
host is "i386-pc-mingw32"
host_alias is "mingw32"
As you can see, the ${host} value is almost always
the "more correct" value to use in this case. There
could be any number of aliases, but one should not
need to deal with that if you just stick to using
$host. Here is a quick patch that should fix things.
cheers
Mo
Index: configure.in
===================================================================
RCS file: /home/cvspublic/apr/configure.in,v
retrieving revision 1.331
diff -u -r1.331 configure.in
--- configure.in 2001/07/08 02:48:53 1.331
+++ configure.in 2001/07/09 06:54:09
@@ -91,7 +91,7 @@
dnl
echo "performing libtool configuration..."
-case "$host_alias" in
+case "$OS" in
*os2*)
# Use a custom-made libtool replacement
echo "using aplibtool"
@@ -1255,7 +1255,7 @@
dnl
dnl BSD/OS (BSDi) needs to use a different include syntax in the Makefiles
dnl
-case "$host_alias" in
+case "$OS" in
*bsdi*)
# Check whether they've installed GNU make
if make --version > /dev/null 2>&1; then
