bad rhug/configure

bad rhug/configure

[Per Bothner]

[-- Attachment #1: Type: text/plain, Size: 621 bytes --]

The rhug/configure has a bug preventing rhug from configuring when
when srcdir=".".  The problem is that dirpart gets set to the empty
string, which causes teh following mkdir "$dirpath/$DEPDIR" to fail.

configure claims to be generated using Autoconf 2.52g.  This bug
appears to also be in the aclocal.m4 in the CVS autoconf, so the
bug isn't rhug per se.  However, aclocal.m4 is auto-generated,
and I don't see where it is generated from. (I don't understand
the autotools that well.)  In any case, I'll report the problem
to the autoconf mailing list.
-- 
	--Per Bothner
per@bothner.com   http://www.bothner.com/per/

[-- Attachment #2: conf.diff --]
[-- Type: text/plain, Size: 492 bytes --]

Index: configure
===================================================================
RCS file: /cvs/rhug/rhug/configure,v
retrieving revision 1.23
diff -u -r1.23 configure
--- configure	2002/02/09 19:17:38	1.23
+++ configure	2002/02/23 00:51:15
@@ -4974,7 +4974,7 @@
   	  /^X\(\/\).*/{ s//\1/; q; }
   	  s/.*/./; q'`
   else
-    dirpart=
+    dirpart=.
   fi
   grep '^DEP_FILES *= *[^ #]' < "$mf" > /dev/null || continue
   # Extract the definition of DEP_FILES from the Makefile without

Re: bad rhug/configure

[Anthony Green]

On Fri, 2002-02-22 at 17:04, Per Bothner wrote:
> The rhug/configure has a bug preventing rhug from configuring when
> when srcdir=".".  The problem is that dirpart gets set to the empty
> string, which causes teh following mkdir "$dirpath/$DEPDIR" to fail.

I've rebuilt everything with autoconf 2.52 and automake 1.5 and it seems
to be working again.

I'm also checking in a change to use --classpath= instead of
-fCLASSPATH=.  So rhug will only build with recent 3.1 era compilers.

There's a bit more to clean up, which I'm doing now.

Also, gcj can't build rhino anymore.  There's a SEGV while building
Interpreter.java.

AG