Jeff Trawick wrote:
On 2/3/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
--- apr/apr-util/branches/0.9.x/configure.in (original)
+++ apr/apr-util/branches/0.9.x/configure.in Fri Feb 3 17:20:52 2006
@@ -186,8 +186,7 @@
fi
if test ! -d "$top_builddir/include/private"; then
- mkdir $top_builddir/include
- mkdir $top_builddir/include/private
+ $mkdir_p $top_builddir/include/private
Curious: righteous cleanup unrelated to rest of commit, or the include
directory could sometimes exist when include/private didn't?
Yes in vpath we make include/private on the fly. (We do this before
autoconf has created include/). I almost wrote this syntax, and then
went back and determined it was a bad example; therefore I adopted the
appropriate syntax for creating expat/xml in the VPATH target tree, and
went ahead and fixed this rather silly example before someone else
tried to cut-n-paste it as I nearly did.
Also, if we're moving to $mkdir_p, can't we just eliminate the test
for pre-existence? (dunno if $mkdir_p is somehow hobbled w.r.t.
normal "mkdir -p")
$mkdir_p is an autoconf result which provides a sh script to replace any
borked mkdir -p, but will expand to mkdir -p when it's know to behave.
We don't verify that 'mkdir -p' behaves with respect to existing dirs,
however, so I don't know that we want to eliminate that test without
addressing pre-existance in both our mkdir.sh replacement and the test
for valid 'mkdir -p' behavior.
Bill
