Re: od question

[Paul Eggert]

Nick Stoughton <address@hidden> writes:

> In coreutils, the od command --help says:
> od --traditional [FILE] [[+]OFFSET [[+]LABEL]]

That has changed since coreutils 5.2.1.
The current od --help says this instead:

Usage: od [OPTION]... [FILE]...
  or:  od [-abcdfilosx]... [FILE] [[+]OFFSET[.][b]]
  or:  od --traditional [OPTION]... [FILE] [[+]OFFSET[.][b] [+][LABEL][.][b]]

> Is this a bug in the code (I think so) or a bug in the usage
> statement (and man page)?

Both.  Here's a slightly mangled copy of the patch that fixed these
and some other compatibility bugs:

<http://www.mail-archive.com/address@hidden/msg02553.html>

(Normally I'd give a GNU URL but I can't reach lists.gnu.org right now.)


> The XSI version of
>
> od [-bcdosx][file] [[+]offset[.][b]]
>
> does not appear to be supported at all. Is this changing?

Yes, the XSI version is now supported, as described in the above patch.

> (i.e. the --traditional flag isn't needed)

That flag is still needed for usages like "od --traditional file 1 2".


> Would you like me to develop a patch?

I would dearly love having someone look over the current
implementation's behavior and conformance.  It is a messy area.  For
example, in responding to your email I just discovered a bug in the
post-5.2.1 changes, and installed the following fix which should
propagate to savannah in a few days.  Other than this patch you can
retrieve the current od implementation via CVS or via the web (e.g.,
<http://savannah.gnu.org/cgi-bin/viewcvs/coreutils/coreutils/src/od.c>).

2004-09-27  Paul Eggert  <address@hidden>

        * src/od.c (format_address_paren): c is optional, so don't output
        it if it's '\0'.

--- old/src/od.c        6 Sep 2004 07:46:43 -0000       1.151
+++ new/src/od.c        28 Sep 2004 01:09:03 -0000      1.152
@@ -1138,7 +1138,8 @@ format_address_paren (uintmax_t address,
 {
   putchar ('(');
   format_address_std (address, ')');
-  putchar (c);
+  if (c)
+    putchar (c);
 }
 
 static void