Date: Sat, 24 Jan 1998 13:19:28 -0800 (PST)
Message-Id: <199801242119.NAA03045@adit.ap.net>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: Robert Hoehne <robert DOT hoehne AT gmx DOT net>
From: Nate Eldredge <eldredge AT ap DOT net>
Subject: Re: RHIDE linker problems - HELP!
Cc: djgpp AT delorie DOT com
Precedence: bulk

At 10:59  1/23/1998 +0100, Robert Hoehne wrote:
>Nate Eldredge wrote :
>> 
>> Robert, this is a very confusing error message. Can that be changed?
>
>Probably I'll add a check if $DJGPP is set, and if not, RHIDE will
>bring up a window, telling it the user.
Okay, but it also happens when the user compiles from the command line. I
meant, should the message issued by the linker be changed, since it is not
very indicative of the actual error.

I looked inside the linker a bit, and it seems the problem is that
`ldfile_open_file' (file ldfile.c) prints the BFD error message when
something goes wrong. But since no BFD functions were called when searching
(it discovered it had no search path), bfd_error is never set appropriately
and its old value is printed, which I believe is a relic of some
format-checking.

Actually, I wonder if GCC should also be made to print an error message if
$DJGPP is not set. Something along the lines of "DJGPP variable not set.
Read the README.1ST file!!!" It could even exit at that point to force the
luser to RTFM before proceeding. This would hopefully cut down on postings
like this, since such a user might be too lazy even to run djverify.

Nate Eldredge
eldredge AT ap DOT net