This is the mail archive of the java@gcc.gnu.org mailing list for the Java project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Testsuites and Test Results


> Ranjit> After digging around a bit, I finally found out that for
> Ranjit> some reason, my runtest from DejaGNU does not accept
> Ranjit> "runtest TCLVAR=VALUE" style invocations and that's what's
> Ranjit> causing these problems!
> 
> I know I've run across this bug before.  I forget what caused it.  I
> had a hacky dejagnu patch to make it work, a long time ago.  I'm not
> sure where you would find discussion of it :-(.  Maybe the java list
> archives.

As David points out elsewhere and as I found out
myself, 1.4.2 only partly has the code in runtest.exp
to parse "TCLVAR=VAL" style command line arguments.

The first pass through the command line accepts it,
but the second pass rejects it. >:-(

The fix is to simply add the following (taken from 1.4.3)
to the second pass (before the "default" case label):
---------------------------- 8< ----------------------------
+       "[A-Z0-9_-.]*=*" { # skip makefile style args like CC=gcc, etc... (processed in first pass)
+               continue
+       }
+
---------------------------- 8< ----------------------------

1.4.3 does not suffer from this and I think that at
least for GCJ, we should make DejaGNU 1.4.3 as the
minimum required version. However, I do not know for
sure whether this version will play nice with all
the other testsuites in GCC.

(As an aside, it supports KFAIL for "Known to Fail" which
could help disambiguate the case with XFAIL as I had
asked earlier.)

Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT hotmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]