Following the corrruption of my USB-stick installation of NetBSD-current amd64 (6.99.44), I used my USB-stick installation of NetBSD-6.99.44 i386, base system with no packages, to build and install NetBSD-current amd64 and i386 on two hard drive partitions (GPT).
Even with 32 MB RAM, building userland and GENERIC kernel, including native X, took from 16:10:00 UTC to 21:26:21 on Mon Aug 11 to build for amd64 and 9:44:18 to 14:06:42 on Aug 12 to buid for i386. I realize that from i386 without PAE, I had effectively 4 GB RAM, not 32 GB. I set up nonroot account and found I could not access man page because file in /tmp could not be created, permission denied, but it worked when I made /home/arlene/mydir and export TMPDIR=/home/arlene/mydir (nonroot user here being arlene). I also could not startx because of permissions problem on /var/log/Xorg.0.log . I never had this particular problem before on any multiuser OS: NetBSD, Linux, FreeBSD, OpenBSD. Has anybody else noticed this? I could startx as root, only with DRMKMS, but mouse was dead (not detected), though I was able to set up a surrogate mouse with x11/xkbset from pkgsrc. I can't copy-and-paste with the mouse on DRMKMS text console, but moving the mouse makes some characters disappear from the screen, so I know the mouse is sort of detected. That was better that what I had with modular (pkgsrc) Xorg, where X woudn't start at all, could not find any screens. I was really expecting to get 6.99.49 or 6.99.50 rather than 7.99.1, though I knew the branching was nigh. Now I could continue to update HEAD (7.99.x) or switch to releng netbsd-7 . I don't want to have to rebuild all packages on amd64 installation now. I suppose releng-7 might be the stabler choice, and more testing is needed from many users before NetBSD-7.0_RELEASE is ready. Tom
