Re: libjava holding regression tester hostage

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Mike" == Mike Stump <mrs@apple.com> writes:

Mike> but, this is wrong, it doesn't have a - on the front of the runtest
Mike> command, so it propagates `normal' testsuite failures all the way up
Mike> to  the return status of make, which, we decided a while ago not to
Mike> do.

It is trivial to fix this in automake[1], though of course that would
mean a new automake release of some kind before we could use it.

But... why is it best to ignore the error in the target Makefile?
Other choices would be the top-level Makefile, or dejagnu itself.

For instance, the top-level Makefile would do what automake-generated
Makefiles do, and ignore subdir 'make' failures when -k is passed.

Tom


[1] I think this patch suffices.

Index: lib/am/dejagnu.am
===================================================================
RCS file: /cvs/automake/automake/lib/am/dejagnu.am,v
retrieving revision 1.23
diff -u -r1.23 dejagnu.am
--- lib/am/dejagnu.am 8 Nov 2003 16:07:45 -0000 1.23
+++ lib/am/dejagnu.am 25 Mar 2005 20:25:18 -0000
@@ -1,5 +1,5 @@
 ## automake - create Makefile.in from Makefile.am
-## Copyright (C) 1994, 1995, 1996, 1997, 2000, 2001, 2003
+## Copyright (C) 1994, 1995, 1996, 1997, 2000, 2001, 2003, 2005
 ## Free Software Foundation, Inc.
 
 ## This program is free software; you can redistribute it and/or modify
@@ -46,7 +46,7 @@
 .PHONY: check-DEJAGNU
 check-DEJAGNU: site.exp
 ## Life is easiest with an absolute srcdir, so do that.
-	srcdir=`$(am__cd) $(srcdir) && pwd`; export srcdir; \
+	-srcdir=`$(am__cd) $(srcdir) && pwd`; export srcdir; \
 	EXPECT=$(EXPECT); export EXPECT; \
 ## Allow this to work when expect and DejaGnu are in tree.
 ## Only required when --cygnus in force.