Re: lib/60530: NetBSD 11.0 cleandir leaves external/cddl/osnet/lib/libdtrace/obj/dtrace.expsym around

The following reply was made to PR lib/60530; it has been noted by GNATS.

From: =?UTF-8?Q?Alex_R=C3=B8nne_Petersen?= <[email protected]>
To: [email protected], [email protected], [email protected],
 [email protected]
Cc: 
Subject: Re: lib/60530: NetBSD 11.0 cleandir leaves
 external/cddl/osnet/lib/libdtrace/obj/dtrace.expsym around
Date: Sun, 02 Aug 2026 16:20:18 +0200

 On Sun, Aug 2, 2026, at 16:15, Martin Husemann via gnats wrote:
 > The following reply was made to PR lib/60530; it has been noted by GNATS.
 >
 > From: Martin Husemann <[email protected]>
 > To: [email protected]
 > Cc: 
 > Subject: Re: lib/60530: NetBSD 11.0 cleandir leaves
 >  external/cddl/osnet/lib/libdtrace/obj/dtrace.expsym around
 > Date: Sun, 2 Aug 2026 16:13:28 +0200
 >
 >  On Sun, Aug 02, 2026 at 02:10:01PM +0000, [email protected] via gnats wrote:
 >  > As of NetBSD 11.0, a cleandir invocation leaves
 >  > external/cddl/osnet/lib/libdtrace/obj/dtrace.expsym around. This is not
 >  > normally a problem, but after e1fa74cf44d45b866247e909658488492a8742dd
 >  > (in the Git mirror), it means that doing a series of NetBSD builds for
 >  > different architectures will eventually run into a diffsym failure for
 >  > i386 and amd64 because that file will have been previously generated
 >  > for a non-x86 architecture.
 > 
 >  (a) cleandir not removing that file is a bug
 >  (b) you can not share object directories for different architectures.
 
 To be clear, I'm always doing cleandir before and after distribution. So, this bug notwithstanding, I would expect there to be no artifacts accidentally shared between architectures. Is that right?