Subject: Re: pkg/32127: Update of CMake from 2.0.5 to 2.2.2
To: None <pkg-manager@netbsd.org, gnats-admin@netbsd.org,>
From: Lars Nordlund <lars.nordlund@hem.utfors.se>
List: pkgsrc-bugs
Date: 11/20/2005 22:56:01
The following reply was made to PR pkg/32127; it has been noted by GNATS.

From: Lars Nordlund <lars.nordlund@hem.utfors.se>
To: "Jeremy C. Reed" <reed@reedmedia.net>
Cc: gnats-bugs@NetBSD.org
Subject: Re: pkg/32127: Update of CMake from 2.0.5 to 2.2.2
Date: Sun, 20 Nov 2005 23:55:01 +0100

 On Sat, 19 Nov 2005 17:35:59 -0800 (PST)
 "Jeremy C. Reed" <reed@reedmedia.net> wrote:
 > > -share/CMake/Modules/Platform/NetBSD.cmake
 > > +share/CMake/Modules/Platform/${OPSYS}.cmake
 > 
 > Please don't make this one line change. See my PLIST revision 1.6:
 > 
 >    "Also hardcode "NetBSD" in PLIST instead of ${OPSYS} so deinstall
 >    works on non-NetBSD platforms."
 
 Ouch. That was the output from 'make print-PLIST' I used. For some
 reason it put ${OPSYS} instead of NetBSD. When is it good to have a
 variable like OPSYS in the PLIST? A generated file which is assumed to
 have the name of the operating system? Seems to me it can break more
 than it automatically fixes? I see that some packages in pkgsrc-wip are
 using it but noting in regular pkgsrc.
 
 How about a 'make print-PLIST' which does the regular things which are
 SAFE, quiet. Things it does which are NOT SAFE, like inserting OPSYS, is
 also noted on stderr? It would have helped me, I guess.
 
 
 Best regards
 	Lars Nordlund