From: Clinton E. <men...@cr...> - 2001-02-02 01:59:22
|
Hello. We had a large delay in starting development... Mostly because all of use have school and / or jobs, so we can't get much done. And, the fact that what we were trying to do was too much, and pointless. We have come up with a new plan for mentalUNIX(which is no MentaLin ux due to possible trademark concerns): 1. Fork from debian to get started Why? This gives us a nice, stable base that has lots of pre-package programs, and most of the work is done. The hard part is moving away from debian and towards repackaging everything in mpk files(the MPKG file format). 2. Write a nice installer The installer will be using a modified Debian boot fs. Of course, we will be writing a nice, easy to use installer with things the other distributions don't have like:(and some the others do have) - CD-R/CD-RW configuration during install - Automatic selection of a kernel that is best for the machine(processor type, SMP/no-SMP) - Built in help system - The ability to choose a pre-selected set of pacakges(home,business,server,etc) - Automatic install(click 2 or 3 buttons, give root a pass, create a user, hit one more button, leave it) - Powerful package selection tools(during the debian fork stages dselect, aptitude, console apt, and stormpkg will be include[stormpkg will only be for the GUI install]) - Games to play while packages are being installed The installer will also come in several flavors, including a gtk-framebuffers version, and X version(will be used it fbdetect fails, and vid card can be detected), a console curses version(if all else fails), and a non-in teractive version(you set up a nice XML file, pop it on the boot disk, boot and go). The cdr/cdrw setup should be a nice addition -- several people I know have had problems with setting up cd burners, and having it all do ne during install is always nice. The automatic install might not be 100% automatic, but it will take about 5 minutes of interaction from the user to setup. 3. Focuse development on mpkg/mconf mpkg will be a very powerful tool, but not how it was originall intended. Instead of having extremely powerful user end tools, it will make the life of the package maintainer easier. Repackaging and entire distribution wi ll be no problem with mpkg ; just have all your source pacakges in one dir, and have an empty dir for output ready. Run one command, and all the packages will be repackaged. Also, setting up the package will be simpler th an with the rest. Mconf is the control file that mpkg files will use to control what their pacakge does. It will read an xml configuration file, and act upon what is in that file. It will mostly be used for installing files. Mconf will be designed so that when the FHS changes, no repackaging is neccesary. Ex: all doc files used to go in /usr/doc, but the FHS now states they should go in /usr/share/doc. With other package managers, you would have to repackage everything..with mpkg you change one line in the global mconf configu ration file(most likely <doc_dir>/usr/share/doc</doc_dir>). No repackaging. 4. Focus on the end user experience The first packages we will repackage will be desktop application. Mentalinux will standardize on the xfce desktop, because it is lightweight and provides most of the features of gnome, but it feels more like CDE(In fact, it is written that way). We may do work to bring a desktop to xfce(modifications to XFTree so it has icons on the desktop, and not just a file manager window). We will also repackage several packages(and add some too) to make the interface more consitent. Still, people who don't want to do it the mentalinux way won't have to, they can choose whatever desktop enviroment(or none at all) during install. XFce has its own WM(XFwm), but it can be used with any WM(I use it with WindowMaker, some people use it with E). So, those are the general targets of mentalinux now. We should be able to get a good, stable release in about a year. ------------------------------- #indrema @ irc.openprojects.net be a part of the revolution. unknown_lamer |