Comments in patch should explain all.  I've gone back through my dbi-users 
backlog and looked at every `perl Makefile.PL -v` log I could find, just to 
be sure about this before I submitted it.  This behavior of the Oracle makefiles 
seems to be consistent with any Oracle >8.0.5.

-- 
Stephen Clouse <[EMAIL PROTECTED]>
Senior Programmer, IQ Coordinator Project Lead
The IQ Group, Inc. <http://www.theiqgroup.com/>
diff -u -r1.1.1.1 Makefile.PL
--- Makefile.PL 2001/12/24 22:55:02     1.1.1.1
+++ Makefile.PL 2001/12/24 23:42:36
@@ -303,6 +303,21 @@
        while ($cmds[0] =~ s/\\$/ /) { # join lines split with \'s
            $cmds[0] .= splice(@cmds,1,1);
        }
+       if (@cmds == 2) {
+               # Newer Oracle (>8.0.5, it seems) seem to disregard ECHODO, in
+               # which case the make output probably looked like this:
+               #      echo foo `bar` baz
+               #      foo shell-expanded-bar baz
+               # So let's check for that, and fix things up if so
+               if (my($foo) = ($cmds[0] =~ /^echo\s+([^`]+)`/)) {
+                       if ($cmds[1] =~ /^$foo/) {
+                               # The second line looks like the expanded
+                               # version of the first; let's fix it up like
+                               # we're expecting it to look and use it
+                               @cmds = ("true echo $cmds[1]");
+                       }
+               }
+       }
        if (@cmds == 1) {
            my $build = $cmds[0];
            $build =~ s/^\s*(true\s+)?(\S+)\s*//;       # remove 'true' and 
compiler/linker

Attachment: msg07776/pgp00000.pgp
Description: PGP signature

Reply via email to