dyad
Group: Members
Posts: 9
Joined: Dec. 2003 |
 |
Posted: Jan. 02 2004,19:10 |
 |
How can DSL shake the Big Linux tree? Let's think about it a bit. There are over 6000 linux software packages that are currently maintained around the world. There are three types of software packaging apt-get, rpm, and tar.gz and some 180 different Linux distributions. That creates a lot of confusion and wasted parallel effort. Of course it is all natural, and organic, nobody used force and made it happen this way. Because rpm is environment sensitive (SuSE or Redhat, etc) it is mostly only functional within commercial company distributions. (If you do rpm you get stuck inside of SuSE.) Tar.gz is a long way form being automated, so it seems to me that Debian packaging will be the ultimate winner in the long run.
The major reason people don't use Debian is because it is so hard to get started. Too many choices, too much stuff to download with a modem, and too hard to boot up and get to a word processor. With SuSE it only takes a few hours, and not a few days.
This is where DSL can really change the Linux world. DSL is really cool and offers a lot that is not found in other distributions, like very small size and very easy to install. But DSL is also like a pin stuck in the wall. Its core is done by one person. That is just too much work to keep updating the kernel and keeping it current.
It seems to me that building a set of bash or perl scrips, that would automate the build process into two stages would solve a lot of problems. Stage one is a minimum base, stage two would be an application base.
The starting base stage should be something around 40 megs which would have live-boot from CD, X-windows, Fluxbox, hardware detect, network basics, rxvt, vi, wget and not much else. It would contain programs start and programs end variables, that direct the transformation from HD build to CD-live-boot build process. It might also have a set of hooks, to allow the application base to glue itself into the base.
The application stage build scripts could then provide web browser choice, text editor, word processor, web server, and other options. Because it is based on the starting base, it could be transformed using the live-CD script into a boot CD, by modifying the programs start and programs end variables.
Besides keeping DSL current, it would also allow for diversity. Which application do you want? An item check list could be created which would allow selection. Use cdrecord or not, along with other configuration issues. The user could pick the right mix for them and their needs.
I also think the name 'damnsmall' should be changed, because DSL would have improved debian's install and configuration process. The live CD would be a perfect rescue disk. I think DSL should be dropped for 'Debian-Live' or a more general purpose label.
DSL has shown what can be done with Debian, it explodes where Linux can fit and shortens the time it takes to get there. If this transformation process can be codified and shared, it will shake the big Linux tree and save lots of effort for years to come.
Maybe several years from now, it will be possible to configure a live R-W CD with choice of processor optimizations, browsers, software tools, and pick a cron job to upgrade it and keep it current. Magic stuff that is easy to use. This could spell the end of SuSE and Redhat.
Just speculating. Not a complaint in any way.. Bill Savoie
|
.
DSL Market ,
Great VPS hosting provided by Tektonic
