> Funny - I never used the "complicated" management tools - just adding and > removing symbolic links was enough for me - for example, I started using > chkconfig from RH7 but now I find rc-update is much simpler than anything > else.
They weren't associated with runlevel management. They were used for associating services (like printers or logins) with serial ports and the like. On BSD you edit /etc/ttys and printcap, and on SVR4 you run pmadm and svcadm with half a dozen obscure arguments which I always had to look up... > > But in general, my impression was that the Bell Labs/AT&T/Lucent > > stuff was always cleaner and more consistent, while BSD was more > > powerful and flexible (because Berkely added features that they > > wanted, even if they violated Unix system philosophy). > > So to restart a service in BSD, I have to find out where it gets launched > from in one of the rc scripts, then look at that file to see if there's > any command-line parameters, and if there are, look in a completely > different file to figure out what the default command-line parameters are > for this one service, then build a command-line manually and run the > service. > > This seems "a tad" more complicated than just doing '/etc/init.d/apache > restart' to me. I think we are arguing on the same side here. Runlevel management is one of the things which I said I thought was done better in SystemV. > Its the little things that tend annoy me with the BSD systems (I manage > BSD and Linux servers for a living but the BSD servers definately require > more effort to manage the simplest of tasks). Both Linux and BSD are re-implementations of Unix, so the annoyances will tend to be 'little things' on the periphery. If you want big things to get annoyed with, that is what Windows is for... > > Linux does a pretty good job of picking the best technology from > > the varous systems, but where it tends to fall down is in tidying > > up the loose ends and producing a coherent product. I suppose that > > is an inevitable result of the more egalitarian development process. > > Also a result of many many more developers working on a faster development > cycle, using the best ideas from other systems/distros/Unices all over the > place. There is no standard way to manage packages or services, even > though there are several de facto standards such as RPM. But those are the > very things that differentiate the various distros of Linux... :-) Oh Definately. And it was dissatisfaction with RPM that was one of the incentives to give gentoo a try. I have my own ideas about how packages should be managed (or at least how I would like them to be managed), but I am not expecting any existing distro to want to drop their existing mechanism to please me. So until I have the time to invent my own distribution, I just have to find the one that is closest to what I want... The sad fact of life is that when you switch to a new OS, you can't help being acutely aware of anything that is done less well in the new system, because you miss what you used to have. But on the converse side you don't appreciate the improvements because you havn't learned to use them yet. That is why it is so easy to become religiously tied to one system. So when I point out a perceived deficiency, it is not because I want to complain. It is because I acknowledge that perhaps there is some good reason for it being the way it is, and I want to give people more familiar with the system an opportunity to persuade me before I conclude that it really is a defect. Regards, DigbyT -- Digby R. S. Tarvin [EMAIL PROTECTED] http://www.digbyt.com -- [email protected] mailing list
