[Midnightbsd-users] MidnightBSD Status Report

[Lucas Holt]

MidnightBSD 0.4-CURRENT is coming along nicely.  We're currently at about 2000 ports working on amd64 and i386 architectures.  The core pieces of Gnome 2.32 are in and several build problems with existing ports have been fixed.   

mport tools are working well now.  They will be the default starting in this release.  pkg_tools will be removed soon from current. 

A new search tool was added to MidnightBSD called msearch. It was designed for full text search of the file system.  It does require significant disk space in /var to operate with large file systems.  The index is compressed with gzip compression.

The largest hurdle for the next release is the installer.  It must be modified to work with the new package tools and that means getting libarchive + libmport built in.  The crunch tools hate liblzma; mport packages use this compression format.  My plan is to make a live cd installer that does not require booting into sysinstall (as init).  I've been researching approaches to this problem. 

Today, I added support for pipes to define  unique st_dev, st_ino value pair for fstat(2).  This is required by sus4.  

Two security issues have come up with MidnightBSD in the last few weeks.  compress and unix sockets were modified to work around them in MidnightBSD 0.3-RELEASE-p3 and later p4 corrects a problem with the unix sockets fix and the linuxolator.  

ath(4) and alc(4) were both updated recently.  I've been testing alc(4) on my new laptop and it's working a lot better now.   Previously, it would not function until the interface went through a down/up cycle and the cable was plugged back in.  ath(4) works with several newer atheros wireless cards.  I'm working on a fairly popular chip now that is included in my new laptop so I can remove linux from the system. :) 

I've been thinking ahead to the 0.5 release as we get closer to 0.4-release. I'm considering the following for 0.5:

1. Changing the system compiler and binutils.  This could be a newer GCC or LLVM.  DragonFly has successful migrated to GCC 4.4, but there's the licensing question with that.  Do we want GPLv3 stuff in base?  
2. Graphical installer
3. Graphical mport package management
4. Final decision on desktop environment. 
5. Updating the objective-c runtime.
6. removing bind from base and just keep the resolver library and a few utilities.  BIND would be available in mports.  This is a bandwidth issue in keeping up to date on DNSSEC and security patches.  

I see three paths for desktop use. 

1. KDE
2. Gnome
3. GNUStep + Etoile (or custom wm + apps)  

PC-BSD has done a good job on the KDE front and I'd feel that is redundant.  I do like KDE 4 and use it at work.  GhostBSD is now doing a very nice installation with Gnome.  Again, I feel a gnome approach might be redundant.  Also, unlike the KDE community, Gnome folks don't' seem to friendly to BSD.  Finally, GNUstep related environment.  This was my original plan and that project has rejected upstream patches.  They've also changed the licensing to GPLv3 since I started MidnightBSD.  I'm left with a few choices including maintaining a patch set for GNUstep or forking it.  Neither sound too appealing, although using a GPLv2 version would be less painful.  

It's just as important to make MidnightBSD unique as it is to make it competitive.  We have limited resources to accomplish this task.   Rather than keep maintaining ports for KDE, Gnome, GNUstep and so forth, I'd like to focus on one of them.  It would speed things up a great deal. We'd still have qt4 and gtk ports regardless.  Without them, we couldn't offer office suites or web browsing.  This is a tough decision and I'd like some input from users and potential users.  


Lucas Holt
Luke at FoolishGames.com
________________________________________________________
MidnightBSD.org (Free OS)
JustJournal.com (Free blogging)