Re: [GNUstep-packagers] patch for observance of $HOME

[Armando Di Cianno]

On 2004-08-06 03:36:51 -0400 Matt Rice <ratmice@yahoo.com> wrote:
> personally I don't think we should trust environment
> variables for reasons in the archive..
> http://lists.gnu.org/archive/html/bug-gnustep/2003-11/msg00015.html

In general, I have to agree.  There does exist some amount of status 
quo usage of "one the command line being run" environment variables, 
especially when building software: configure, make, etc, can often 
change behavior based on a env var you pass in "just that once".

> and would rather if we did something like cleanse
> make/GNUstep.sh of writes to the user root, so that it
> isn't breaking the sandbox if possible, so he doesn't
> have to redirect $HOME to a directory inside the
> sandbox (if i grok)

Yes, you grok. :-)  The patch I originally sent isn't too terribly 
hackish as it tries to observe $HOME in NSUserHome(), which at least 
makes sense.  I say it that way, as I was really trying to avoid 
"fixing the symptoms" of a GNUstep.sh that needs to be overhauled.  I 
also have a patch that breaks up GNUstep.sh into GNUstep-env.sh and 
GNUstep-user-update.sh (with GNUstep.sh calling both for backwards 
compataibility).  Building docs still bombed out, but I got all the 
way to gnustep-gui before the GNUmakefiles in there starting calling 
various tools that tried writing defaults.