Gnome port

[Conrad Sabatier]

On 05-Feb-2006 Christopher Sean Hilton wrote:
> Is there a reason that the gnome 2 ports don't use the conflicts
> mechanism to avoid completely hosing an existing gnome 2 install? On
> Friday I came across a gimp script-fu which would slice an image into
> pieces and output html that rendered the complete image as a
> table. But it was written in python and my gimp doesn't understand
> python-fu. Well I figured that it would be a simple matter to:
> 
>      # cd /usr/ports/graphics/gimp; make config
> 
> and tell the the port that I wanted to use python plugins. And then a
> simple:
> 
>      # make
> 
> would either build the package or _harmlessly_ barf trying. Well, it
> did barf but only after it had installed gnome2-vfs-2.12.x.x or
> sumsuch and about 5 or six other gnome2-2.12 ports on top of my
> working gnome2-2.10.x.x installation. Obviously this rendered the
> installed gnome2-2.10 ports on my box inoperable in very difficult to
> debug ways.
> 
> Would a CONFLICTS entry in the apropriate make files have stopped the
> gnome overbuilds and left my system in a working state?

I believe you might have avoided this sort of problem if you had used
portinstall (portupgrade) to add the new package.  In that case,
portupgrade would have detected that newer versions of some of your
existing GNOME ports were available, and rather than simply installing the
new versions on top of the old, it would have upgraded the already installed
ports to the newer versions.

> Also please don't get me wrong. I planned to update my laptop from
> 5.4-STABLE to 6.0-STABLE later this week.

That's really beside the point in this case.  The same thing could have
happened regardless of which FreeBSD version you were running at the time.

> But this problem means that
> I'm without my laptop for a client visit and mailserver build which is
> not making me very happy.

Sorry to hear that.  But we've all made our share of these types of
mistakes on the path to greater knowledge, wisdom and understanding.  :-)
 
> Thanks
> -- Chris
> 
> P.S. I glad to post a bug report an fling this email in the direction
> of the port maintainers. Heck I think I could even script the process
> of updating the Makefiles given enough time.

I'm not sure if the behavior you witnessed could rightly be called a "bug",
but it certainly could be considered an issue worth addressing.  Obviously,
running "make install" instead of using portupgrade can be a risky
proposition in some cases, and at the very least, the handbook should make
this clear.

-- 
Conrad J. Sabatier <conrads at cox.net> -- "In Unix veritas"