[Freeciv-Dev] [MAILER-DAEMON@xxxxxxxxxxxxxxxxx: Returned mail: freeciv-smtp-03.complete.org. config error: mail loops back to me (MX problem?)]

[Greg Wooledge <wooledge@xxxxxxxxxxx>]

We'll try this again....

----- Forwarded message from Mail Delivery Subsystem 
<MAILER-DAEMON@xxxxxxxxxxxxxxxxx> -----

Date: Mon, 2 Aug 1999 22:29:35 -0500
From: Mail Delivery Subsystem <MAILER-DAEMON@xxxxxxxxxxxxxxxxx>
To: <wooledge@xxxxxxxxxxx>
Subject: Returned mail: freeciv-smtp-03.complete.org. config error: mail loops 
back to me (MX problem?)
Auto-Submitted: auto-generated (failure)

The original message was received at Mon, 2 Aug 1999 22:25:17 -0500
from freeciv.freeciv.org [206.251.228.174]

   ----- The following addresses had permanent fatal errors -----
<freeciv-dev@xxxxxxxxxxx>

   ----- Transcript of session follows -----
553 freeciv-smtp-03.complete.org. config error: mail loops back to me (MX 
problem?)
<freeciv-dev@xxxxxxxxxxx>... Deferred: No route to host

Reporting-MTA: dns; kiss.otherbbs.com
Received-From-MTA: DNS; freeciv.freeciv.org
Arrival-Date: Mon, 2 Aug 1999 22:25:17 -0500

Final-Recipient: RFC822; freeciv-dev@xxxxxxxxxxx
Action: failed
Status: 4.4.1
Remote-MTA: DNS; freeciv-smtp-04.complete.org
Last-Attempt-Date: Mon, 2 Aug 1999 22:29:35 -0500

From: wooledge@xxxxxxxxxxx
Date: Mon, 2 Aug 1999 20:10:13 -0700 (PDT)
To: freeciv-dev@xxxxxxxxxxx
Subject: Re: [Freeciv-Dev] CVS compile problem with i18n (PR#92)
CC: bugs@xxxxxxxxxxxxxxxxxxx

David Pfitzner (dwp@xxxxxxxxxxxxxx) wrote:

> I've seen this problem when first doing ./configure --with-included-gettext,
> and then later configuring normally (when configure finds and uses the
> system gettext).  The problem is that the first method (and I think
> also configure --disable-nls) creates a symlink in freeciv/intl/ of
> libintl.h to libgettext.h.

Ah.  I saw the definitions in there, but didn't realize that that
symlink was conditional.

I ran ./configure, then observed that I didn't have gettext.  Something
blew up (the Makefile tried to run a "no" command -- details below), and
then I installed Debian's gettext package.  Running "rm config.cache"
and "./configure" wasn't sufficient, it seems.

> If the symlink _is_ there, a solution is simply to remove it.

Done.  Fixed.  Thanks. ;-)

> But I'm not sure what a general solution to avoid this problem 
> in the first place might be... 

A brute-force approach might be to remove intl/libintl.h at the beginning
of configure.  It will be created at the end of the configure process
if it's needed, right?  (*Slightly* more elegant might be to remove it
if config.cache is absent.  That way, people like me who *almost* know
what we're doing and remove config.cache will also get intl/libintl.h
removed for us....  But that's not much better than just removing it in
all cases, I think.)

Back to that "no" command....  I'm guessing that this comes from one of
those configure lines -- you know, the ones that say stuff like

checking for main in -lnls... (cached) no

This probably put the value of "no" into a shell variable, which was
then propagated into a Makefile somewhere.

I figured I could install gettext, rerun configure and make, and it
would pick up pretty close to where it left off -- then I'd see what
command the "no" was supposed to be, and use that as a jumping-off point.
Unfortunately, it didn't work out that way, as we can see from this
bug report.

-- 
Greg Wooledge                    | Distributed.NET http://www.distributed.net/
wooledge@xxxxxxxxxxx             | because a CPU is a terrible thing to waste.
http://www.kellnet.com/wooledge/ |



----- End forwarded message -----

-- 
Greg Wooledge                    | Distributed.NET http://www.distributed.net/
wooledge@xxxxxxxxxxx             | because a CPU is a terrible thing to waste.
http://www.kellnet.com/wooledge/ |