IIRC, Jared mentioned that the lang/rust was problematic, consuming large amounts of ram, and then ended up being hit or miss. It is possible to build, but changes of success are not deterministic.
Obviously, this has implications for dependent packages (eg, mate, which is how I found it). It looks to have had issues in 2019Q3 as well. (Or maybe because I was trying to build on a Pinebook. ;) )
On 26 Dec 2019, at 13:37, Greg Troxel wrote:
I am curious if the major failures below, specifically
lang/g95 515 wennmach%NetBSD.org@localhost
devel/ocaml-findlib 194 jaapb%NetBSD.org@localhost
lang/go14 183 bsiegert%NetBSD.org@localhost
lang/rust 161 pkgsrc-users%NetBSD.org@localhost
misc/raspberrypi-userland 133 jmcneill%NetBSD.org@localhost
devel/kio 127 markd%NetBSD.org@localhost
www/webkit-gtk 48 pkgsrc-users%NetBSD.org@localhostare really broken for earmv7hf, or if they are bulk build artifacts.
Perhaps more to the point, I am curious if they seem fixable for the
freeze, i.e. in the next day or so, or not.