Re: pkg/44597 (add native Kerberos support on Solaris)

[Thomas Klausner <wiz%NetBSD.org@localhost>]

The following reply was made to PR pkg/44597; it has been noted by GNATS.

From: Thomas Klausner <wiz%NetBSD.org@localhost>
To: NetBSD bugtracking <gnats-bugs%NetBSD.org@localhost>
Cc: 
Subject: Re: pkg/44597 (add native Kerberos support on Solaris)
Date: Tue, 30 Aug 2011 16:56:47 +0200

 On Tue, Aug 30, 2011 at 01:15:06PM +0000, Tim Zingelman wrote:
 >  Other than the fact that mit-krb5 (non-native) IS 1.8?
 
 That alone is never a reason :)
 
 >  (which was both an API & ABI bump from the previous pkgsrc version as
 >  far as I understand the definition of those variables)
 
 We bump ABI if a package compiled against that version won't compile
 against an older version. One example is a shlib major bump.
 Then packages built against it are linked against libfoo.2 while the
 old version would only provide libfoo.1 and the old package wouldn't
 fulfill the new dependency.
 
 Users can override this by setting ABI_DEPENDS to "no" but then they
 are on their own. (You might want to do this if e.g. glib2 is bumped
 and you don't have time to reinstall all packages, but just want to
 update one of the packages using it.)
 
 We almost never bump API.
 One reason would be if all packages using it NEED the newer version.
 I can't think of the other reasons right now, but I think there were
 more.
 
 In packages using a package which need a particular version (e.g.
 because of added functions), you would override API.
 
 >  Why is it that those variables are used both to say what version is
 >  provided by a package
 
 The version provided by the package is the PKGVERSION.
 
 
 >  Are these variables better explained/defined someplace other than
 >  http://www.netbsd.org/docs/pkgsrc/buildlink.html ?
 
 Not sure. Feel free to add the explanation, or I might come around to
 it :)
  Thomas