It ain't broke, don't fix it!

More and more often, I get the occasion for a violent rant. This means one of two things: either a) I become older and crazier or b) this world becomes too stupid and too crazy for me.

Being one of those happy beings that never manage to get enough time and which think sleeping and eating are a huge waste of time, I, normally, conform to all sorts of Murphy laws, get a nasty time wasting from time to time.

This evening, I planned to prepare the two laptops granted to KDE for the next show I go to. I won't mention the very kind sponsor and the show's name here, in the context of my bad rant. I will come to them and the deserved praise in due time.

Well, while planning those laptops, I decided to give a try to SuSE-10.1. Good occasion to showcase to the future visitors of the KDE booth the latest in KDE and Linux technology. While all the installation went smooth and things look polished and well thought out, there is one smelly turd in the flower garden: the new updating tool.

I usually stay away from speaking (good or bad) of commercial technologies and companies that promote them, especially when I can be accused of vested interest. But this time, I need to say it loud. Three hours lost of my almost non-existent free time warrant me the right to do it. Here it is:

Novell, if it ain't broken, don't fix it. Yast's update tools in SuSE up to 9.3 worked marvelously. I use it very often and I can struggle very hard to remember the last time I had a problem. The zenworks thingie that is meant to replace yast-update-online is simply not working. No way to add new repositories, updates die in the middle, total mess. Talk of over-engineering. Does really an updating system need a server-client architecture, with its own 200-command vocabulary, with daemons that need 221s to go from "sleep" mode to active mode?

Be sensible and either fix it, or give me back y-o-u.


...why they did it: first they wanted to merge two solutions they had: read carpet and yast. Since it is bad to displease a set of developers of one solution it makes sense to merge them.
Additionally, afaik both solutions do not come along with a real mirror solution - remember the release of Suse 10.0, where all the big mirrors broke down? This would not have happened if there would have been a real mirror solution (yes, I know about 302', but it didn't work obviously).

Not to mention that there is slowly but steadily a standard repository format for rpm packages comming up (XML package metadata) which is used by more and more tools, and the new solution should be able to deal with that, too (in case the new format becomes kind of a de-facto standard for LSB or something).

Anyway, if I would use Suse on a daily base, I would use yum - that's the solution of Fedora Core/CentOS which *is* supported by Suse (at least in the main repositories). A set-up howto can be found here.



By liquidat at Sat, 05/20/2006 - 13:36

Neither from points of view commercial nor in a FOSS perspective. Commercially, if trying to not displease a hand of developers means alienating a large number of your potential clients, then you just bite the bullet and displease them. In FOSS, the ecology of FOSS projects plays it all. The best fitted stays. In this case, it would have been yast, which made its proof for many years.

I _bought_ SuSE packs since 5.X series (cca 1997, I think). I can't convince myself doing it again if this means having broken tools forced onto me.

A pitty for the excellent work of the distro designers and the KDE team, which will take a hit because of something they don't control.

By Cristian Tibirna at Sat, 05/20/2006 - 15:15