On Mon, 16 Aug 2004, Michael L. Hitch wrote:

>   A quick workaround is to do a make patch in lang/perl58 and edit
> the MANIFEST file to change the " NaN ", then continue building.

  That didn't work so well - the perl build does two builds: one with
shared libraries and one static.  The static build unpacked and patched
the perl distribution and awk croaked again on it.

  A different workaround is to provide a patch to adjust the MANIFEST
entry causing the problem:

patches/patch-zd:

--- MANIFEST.orig	2004-08-17 06:04:37.000000000 -0600
+++ MANIFEST	2004-08-17 06:02:23.000000000 -0600
@@ -1373,7 +1373,7 @@
 lib/Math/BigInt/t/downgrade.t	Test if use Math::BigInt(); under downgrade works
 lib/Math/BigInt/t/_e_math.t	Helper routine in BigFloat for _e math
 lib/Math/BigInt/t/fallback.t	Test Math::BigInt
-lib/Math/BigInt/t/inf_nan.t	Special tests for inf and NaN handling
+lib/Math/BigInt/t/inf_nan.t	Special tests for inf and *NaN* handling
 lib/Math/BigInt/t/isa.t		Test for Math::BigInt inheritance
 lib/Math/BigInt/t/mbf_ali.t	Tests for BigFloat
 lib/Math/BigInt/t/mbi_ali.t	Tests for BigInt


distinfo entry for patch-zd:

SHA1 (patch-zd) = b73918fe1ee892ec0043bd39b5db4fcac4658cf5

  My build of mit-krb5 is currently doing the second build of perl;

...waiting until Configure gets to the awk command again...

  And it got past the awk command again, so the patch gets around the
problem - at least until something else feed "NaN" to awk.

  The real fix is to deal with the reserved operand fault better.

--
Michael L. Hitch			mhitch@montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA