Re: [bug-idutils] A couple of issues building idutils 4.5 on HP-UX 11

[Jim Meyering]

Richard Lloyd wrote:
> I'm using HP's ANSI C compiler on PA-RISC and Itanium platforms running
> HP-UX 11.11, 11.23 and 11.31 and came across a couple of code issues that
> needed fixing:
>
> * Line 666 of lib/stdio.in.h reads:
>
>   _GL_CXXALIAS_RPL (printf, printf, int, (const char *format, ...));
>
>   which has 4 parameters to the macro instead of 3. The first printf should
>   be removed so that it reads:
>
>   _GL_CXXALIAS_RPL (printf, int, (const char *format, ...));

Thank you for the report.

Here's a patch for that one, in your name.
I'll push it to the gnulib repository later today.

>From 86ba51d2c17219b82b977da0a912e05a3d9bf858 Mon Sep 17 00:00:00 2001
From: Richard Lloyd <address@hidden>
Date: Sun, 20 Jun 2010 13:31:40 +0200
Subject: [PATCH] stdio.in.h: fix compilation failure when using HP-UX 11's C 
compiler

* lib/stdio.in.h: Remove excess _GL_CXXALIAS_RPL macro argument.
This macro takes 3 arguments, not 4.
---
 ChangeLog      |    6 ++++++
 lib/stdio.in.h |    2 +-
 2 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 8fe5099..27099b8 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,9 @@
+2010-06-20  Richard Lloyd  <address@hidden>
+
+       stdio.in.h: fix compilation failure when using HP-UX 11's C compiler
+       * lib/stdio.in.h: Remove excess _GL_CXXALIAS_RPL macro argument.
+       This macro takes 3 arguments, not 4.
+
 2010-06-15  Giuseppe Scrivano  <address@hidden>

        ipv6: fix detection under mingw
diff --git a/lib/stdio.in.h b/lib/stdio.in.h
index ca8960e..135b084 100644
--- a/lib/stdio.in.h
+++ b/lib/stdio.in.h
@@ -661,7 +661,7 @@ _GL_FUNCDECL_RPL (printf, int,
                   (const char *format, ...)
                   __attribute__ ((__format__ (__printf__, 1, 2)))
                   _GL_ARG_NONNULL ((1)));
-_GL_CXXALIAS_RPL (printf, printf, int, (const char *format, ...));
+_GL_CXXALIAS_RPL (printf, int, (const char *format, ...));
 #  endif
 #  define GNULIB_overrides_printf 1
 # else
-- 
1.7.1.664.gf1fb2

> * Lines 150-153 of gnulib-tests/setenv.c read:
>
>       new_environ =
>         (char **) (last_environ == NULL
>                    ? malloc ((size + 2) * sizeof (char *))
>                    : realloc (last_environ, (size + 2) * sizeof (char *)));
>
>   malloc() and realloc() both return (void *), so casting a (void *) to
>   (char **) upsets the HP C compiler. The solution is to cast the
>   malloc()/realloc() return value to (char *) first:

Can we get away with using no cast whatsoever?
i.e., removing the (char **) cast rather than adding two new ones?

>       new_environ =
>         (char **) (last_environ == NULL
>                    ? (char *)malloc ((size + 2) * sizeof (char *))
>                    : (char *)realloc (last_environ, (size + 2) * sizeof (char 
> *)));
>
> On a non-coding issue, someone might want to update the top-level README file
> of idutils, because it makes references to 'mkid' throughout (and no mention
> of idutils at all!) and claims that "Version 4 will follow in the coming
> months"...

Thanks!  I've updated it.