Re: ghostscript-agpl build failure on Linux

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

On Sun, Apr 13, 2025 at 10:30:36PM +0000, nia wrote:
> On Sun, Apr 13, 2025 at 08:39:05PM +0200, Thomas Klausner wrote:
> > I think there should be a recursive bump for the packages using mit-krb5/heimdal.
> 
> I really don't like the idea of pretending there's an ABI break on every 
> platform for a configuration change that only affects some platforms.

I understand, but we don't have a framework to differentiate this kind
of changes per-platform.

> There's no need to waste energy rebuilding everything on NetBSD, for
> instance. Needless recursive revbumps are the opposite of user-friendly.

I understand this as well. On NetBSD, there is no need for this, but
on everything but FreeBSD & NetBSD, there is.

The change you committed leads to different binary packages from
before. In this case, we always bump the PKGREVISION because
otherwise, how can we tell which version of the program we have? That
just leads to inconsistent binary package sets and fallout like the
one reported in this thread, and more.

I think this doesn't mean we shouldn't bump the PKGREVISION, just that
we should think twice before changing global settings like the default
kerberos implementation.

So please do the recursive bump now.
 Thomas