From: sof <unknown>
Date: Tue, 26 Feb 2002 05:03:27 +0000 (+0000)
Subject: [project @ 2002-02-26 05:03:27 by sof]
X-Git-Tag: Approx_11550_changesets_converted~2352
X-Git-Url: http://git.megacz.com/?a=commitdiff_plain;h=bcfa1e5a52b9ce1d38d2217e8cef28c3d0a4ee3b;p=ghc-hetmet.git

[project @ 2002-02-26 05:03:27 by sof]
INFO_VECTOR (MINIINTERPRETER case): gcc-2.96 doesn't
like a mixture of labelled and non-labelled members
in a struct initializer, which is reasonable. So,
be explicit that we're initializing the 'vector' member.

I'd be surprised if this change causes breakage with earlier
GCCen.

A comment next to INFO_VECTOR as to why it has to be empty
in the non-MINIINTERPRETER case, would be nice.
---

diff --git a/ghc/includes/InfoMacros.h b/ghc/includes/InfoMacros.h
index 0c1c593..b181ce5 100644
--- a/ghc/includes/InfoMacros.h
+++ b/ghc/includes/InfoMacros.h
@@ -1,5 +1,5 @@
 /* ----------------------------------------------------------------------------
- * $Id: InfoMacros.h,v 1.17 2001/09/17 22:46:59 ken Exp $
+ * $Id: InfoMacros.h,v 1.18 2002/02/26 05:03:27 sof Exp $
  * 
  * (c) The GHC Team, 1998-1999
  *
@@ -35,8 +35,9 @@
 #define PROF_INFO(type_str, desc_str)
 #endif
 
+/* Why no empty array initializer in the 'else' branch? sof 2/02 */
 #ifdef USE_MINIINTERPRETER
-#define INIT_VECTOR {}
+#define INIT_VECTOR vector : {}
 #else
 #define INIT_VECTOR
 #endif