math/octave bug

[Jason Bacon <bacon4000%gmail.com@localhost>]

There's one remaining bug in octave on CentOS Linux that's described here:

https://mail-index.netbsd.org/pkgsrc-users/2014/03/24/msg019451.html

I don't actually see a patch for Makefile.in in Iain's post, but I think 
his suggestion of making the charset.alias build unconditional is the 
best approach.  I thought about doing a conditional PLIST entry, but 
this could get hairy and require a lot of maintenance to keep in sync 
with the logic in Makefile.in.

I attached a proposed patch to make building charset.alias 
unconditional, as Iain suggested.

If anyone thinks there's a better solution, please chime in.

    JB

--
Earth is a beta site.

$NetBSD$

--- libgnu/Makefile.in.orig	2016-11-13 15:23:00.000000000 +0000
+++ libgnu/Makefile.in
@@ -2690,18 +2690,7 @@ getopt.h: getopt.in.h $(top_builddir)/co
 all-local: charset.alias ref-add.sed ref-del.sed
 install-exec-local: install-exec-localcharset
 install-exec-localcharset: all-local
-	if test $(GLIBC21) = no; then \
-	  case '$(host_os)' in \
-	    darwin[56]*) \
-	      need_charset_alias=true ;; \
-	    darwin* | cygwin* | mingw* | pw32* | cegcc*) \
-	      need_charset_alias=false ;; \
-	    *) \
-	      need_charset_alias=true ;; \
-	  esac ; \
-	else \
-	  need_charset_alias=false ; \
-	fi ; \
+	need_charset_alias=true ; \
 	if $$need_charset_alias; then \
 	  $(mkinstalldirs) $(DESTDIR)$(libdir) ; \
 	fi ; \