Why put the message here, now? It is not we (FreeBSD) who will deprecated it in a certain amount of time. It is nwtime.org that will. Should we not try to upstream this by submitting a patch to them first? Before we apply this patch here?

Mar 19 2026, 3:44 AM

Mar 12 2026

cy abandoned D55823: krb5: Use GNU strerror_r() when building on Linux.

Update to the latest 15-STABLE is all that is needed.

Mar 12 2026, 7:58 PM

cy added a comment to D55823: krb5: Use GNU strerror_r() when building on Linux.

I think this needs a little more investigation because k5-platform.h already includes autoconf.h. jrtc27@'s October patch should have been enough. I will rebuild again (cleanly).

Mar 12 2026, 2:51 PM

cy added a comment to D55823: krb5: Use GNU strerror_r() when building on Linux.

In D55823#1277119, @emaste wrote:

I find it quite surprising that this would just be broken in the upstream build; is it possible this is an artifact of our bespoke build process? Presumably the Implement the POSIX strerror_r API in terms of the GNU strerror_r section has to be used on GNU systems?

Mar 12 2026, 2:20 PM

cy added a comment to D55823: krb5: Use GNU strerror_r() when building on Linux.

In D55823#1277079, @michaelo wrote:

Is this something which should be upstream?

Mar 12 2026, 12:08 PM

cy added a reviewer for D55823: krb5: Use GNU strerror_r() when building on Linux: krb5.

Mar 12 2026, 3:29 AM

cy requested review of D55823: krb5: Use GNU strerror_r() when building on Linux.

Mar 12 2026, 3:28 AM

Mar 11 2026

cy committed R11:f477358549e1: security/nmap: Restore missed ./configure patch (authored by cy).

security/nmap: Restore missed ./configure patch

Mar 11 2026, 9:43 PM

cy committed rGfdddd0058e34: krb5: Fix -Wint-conversion when bootstrapping on GNU/Linux (authored by jrtc27).

krb5: Fix -Wint-conversion when bootstrapping on GNU/Linux

Mar 11 2026, 5:37 PM

cy committed rGb373cf4ff140: krb5: Include <features.h> on Linux so __GLIBC__ can be checked (authored by jrtc27).

krb5: Include <features.h> on Linux so __GLIBC__ can be checked

Mar 11 2026, 5:37 PM

Mar 6 2026

cy abandoned D55678: krb5, kerberos5: Remove no-undefined linker option.

This patch only masks another problem. When this patch is applied, usr.bin/ldd32 fails on the same error. The symbols that are exported in lib/libc.so, are not exported in lib32/libc.so. The problem is elsewhere but not here.

Mar 6 2026, 4:20 AM

cy updated the summary of D55678: krb5, kerberos5: Remove no-undefined linker option.

Mar 6 2026, 12:30 AM

cy added a comment to D55678: krb5, kerberos5: Remove no-undefined linker option.

My Heimdal builds also fail with the same error. Something has recently changed in the tree to cause these two unrelated Kerberosen to fail with the same libc objects not found.

Mar 6 2026, 12:28 AM

cy retitled D55678: krb5, kerberos5: Remove no-undefined linker option from krb5: Remove no-undefined linker option to krb5, kerberos5: Remove no-undefined linker option.

Mar 6 2026, 12:27 AM

cy added a comment to D55678: krb5, kerberos5: Remove no-undefined linker option.

In D55678#1274214, @emaste wrote:

krb5 build failures have been occurring for a while afaik, e.g. PR293258? 9cea0f0be79d should be a no-op since the now-removed (portions of) if conditions were always true .if ${${X_}LINKER_TYPE} != "bfd" || ${${X_}LINKER_VERSION} > 21750 and the part of .if ${${X_}LINKER_TYPE} == "bfd" && ${${X_}LINKER_VERSION} > 21750 after &&

Mar 6 2026, 12:18 AM