After testing on pointyhat and some valuable Makefile suggestions from ru@ (of course approval/review from my mentor, delphij@), I commit ncurses wide character support into 7-CURRENT.
Next, I’m going to prepare 6.x patchset for both ncurses 5.6 update and wide character support.
As requested by ache@, I removed them. The real libmytinfo exists in 3.x and below. After that, it is only a link and libncurses has its functionality. I just got krismail about 1x ports are broken. I fix 4 of them and I don’t think the rest are really broken (they only checks libmytinfo in configure). I’m building a new HEAD jail in my tinderbox, so I can test them.
I should really go to bed (it’s 5am here). 
Half hours ago, I received a mail from core-secretary@ about my src bit. Now I’m punished again for my recent ncurses work. I would like to thank delphij@ again for mentoring me (again). Hope my collection of pointyhats will not grow too fast.
Today, I spent some time to get wide character support in base’s ncurses. Basically, most work were some six months ago. I managed to get lib32 build work by defining WORLD32 in ${LIB32MAKE} and do not set ${INCS} if WORLD32 is defined. So, it looks like
.if !defined(WORLD32)
INCS= ${HEADERS} ${SRCHDRS}
INCSLINKS= curses.h ${INCLUDEDIR}/ncurses.h
.endif
I’m still running a make universe to make sure everything work. At least, amd64+lib32 and i386 build fine. On my i386 current, mutt works flawlessly with ncursesw. I will post the patch to -current once make universe is finished.
After almost four years, the base’s ncurses is finally updated to 5.6 which is released last Dec. This work starts half year ago and restarted few weeks ago. I would like to thank everyone who helped me to do this update, especially delphij@ (reviewing patch and src commit approval) and kris@ (an exp run on pointyhat).
This is my first src commit and there are few errors: I accidentally import bash.core (but cvs automatically ignores it. Thanks God!), a typo, and I forget to do a force commit before removing old ncurses build glue. Well, the best thing is buildworld is ok and the ncurses is updated.
So, the next step is to enable wide character support and merge ncurses 5.6 into RELENG_6.
Today, I removed chinese/xcin (there is a URL explains the reason in the commit message). For some unknown reason, Tools/script/rmport does not show me that there are 3 ports depend on it. After the commit, I am kicked by portsnap index builder. Then, few hours later, after dinner and some TV series, I notice that vd@ disconnects few ports from chinese/ that makes me nervous. So, I take a look and fix them. Anyway, a shiny pointyhat is added to my collections.
Following DragonFly BSD Digest, I noticed there are some work on so called ‘VKernel’, which is similar to User Mode Linux. It looks interesting.
On a RELENG_6 system, it seems that gpt partition is invisible until a ‘gpt show’ command is issued. So, currently, I have to put ‘gpt show da0’ in rc.early (yes, that’s dirty).
After checking geom_gpt.c, it seems that its start function does nothing on RELENG_6. But, in -CURRENT, it does something. So, my guess is on -CURRENT it is fine. Need to ask some experts.
Recently, I bought a ThinkPad X60 with Intel Core 2 Duo CPU. Though, I have an external USB CD-Rom, I decided to try PXE installation. I know this long time ago, but never try it out.
In short, you need a DHCP+NFS+tftp server to start. All the tricks are documented here. If you have NAT running on the server, disable it. Otherwise, some funny things happen. PXE installation is not that hard as I thought. Speaking of PXE, I think there is a project about PXE bundle installer. Hope we can see it soon 