Re: gstep-make, FORCE_USER_ROOT

[Helge Hess]

On Thu, 4 Apr 2002 16:34:06 +0100
"Richard Frith-Macdonald" <richard@brainstorm.co.uk> wrote:
> > Isn't calling user_home unnecessary if I configured with 
> > --with-user-root ?
> 
> Well, user_home tells GNUstep.sh the value you configured.

Yepp, I know. But GNUstep.sh is itself a "configured" file and can use those 
autoconf variables directly ?!
Anyway, this isn't important for me ;-)

> > Problem 2: I'm almost sure now that user_home injects a carriage return 
> > (^M) into my environments variables. Is there anything known which 
> > could  be the reason for this ?
> Nothing I know of.

I still have not found an explanation. Everything was fine till I installed 
make-1.3.0. To make it even stranger it works in the development environment 
just fine, only on our RPM build machine (which is running a pretty old Linux 
in constrast to the dev-machines) it breaks.

> Why not do 'user_home user | od -c' to see exactly what user_home is 
> producing?

I inserted a small loop into user_home.c which checks for ^M before printing 
out the value. And indeed the value does contain the CR :-(
BTW: Sometimes the -L flags seems to be affected too. Something really strange 
is going on ...

> I it *is* producing a carriage return, you need to look at how the 
> FORCE_USER_ROOT
> variable is being defined when user_home is compiled.

That's the next thing I will do.

Thanks,
  Helge
-- 
SKYRIX Software AG    - http://www.skyrix.com
German XML-RPC Site   - http://www.xml-rpc.de