Re: [Jikesrvm-core] Problems with i686-pc-linux-gnu and sanity tests

[Steven A. <au...@us...>]

Oh, how gross.  I get it.  There's a function in jconfigure, 
unsetEnvironment(), that I wrote just to unset the affected variables 
before we 
read in a different config file.

I'm adding more variables for it to unset; I'll check in the change soon; 
tonight's PPC/Linux run should go fine.

--Steve









Chris Hoffmann <hof...@cs...>
Sent by: jik...@ww...
10/21/2003 09:43 AM
Please respond to jikesrvm-core
 
        To:     jikes-core <jik...@ww...>
        cc: 
        Subject:        [Jikesrvm-core] Problems with i686-pc-linux-gnu 
and sanity tests


A few days ago i686-pc-linux-gnu was changed so that it doesn't reassign
HOST_JAVA_HOME if it is already set. Unfortunately, this breaks my
regression tests if the paths to Java are different on the host and
target platforms:

Section 3 of jconfigure reads the RVM_TARGET_CONFIG file, which sets
HOST_JAVA_HOME, as does section 6. Only at the end of section 6 is the
RVM_HOST_CONFIG file read and environment.host written. With the current
version of i686-pc-linux-gnu the HOST_JAVA_HOME is not overriden and the
*target's* java home is written to environment.host.

Also, in function doBuild of script RunSanityTests,
$RVM_BUILD/environment.host is sourced immediately before calling
jconfigure. If things are working properly, environment.host shouldn't
exist yet, no? It's created by jconfigure. As it turns out I don't think
this is causing a problem, but it is a bit odd....

Chris
-- 
Chris Hoffmann -- Dept. of Computer Science/UMass at Amherst
http://www-ali.cs.umass.edu/~hoffmann
_______________________________________________
Jikesrvm-core mailing list
Jik...@os...
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jikesrvm-core