On Wed, 2005-05-04 at 17:26 +0100, Ciaran McCreesh wrote: > On Wed, 4 May 2005 18:17:53 +0200 Michiel de Bruijne > <[EMAIL PROTECTED]> wrote: > | Second showstopper is lack of (LSB-)certification; > | Like it or not, the typical managers in this world say if product x > | isn't y-certified we wont use/support it. Gentoo needs a profile for > | a LSB-compliant system. > > Don't be silly. >
Forgive me to come into a list where I should probably just be lurking, but this is really what I subscribed to the list for: why is this idea silly? Let me give you some background. I used Red Hat for years without fuss, until they pulled their "stunt" around 8.0. (The quality obviously took a hit, and I could tell for sure by 9 what they were planning on doing, even before they announced Fedora.) I threw everything up in the air and landed on SuSE, and have been happy there for a few years. I love the direction Novell seems to be going, but I keep finding show-stopping bugs in their stuff. The latest problem is a hard lock when syncing my new Treo. So I'm trying Gentoo, and really liking it. The problem I'm having with Gentoo is the whole "build" thing. I know, I know. That's the whole point. But I'm trying to bring this thing up on an old PIII laptop with 256 MB of RAM. Even though this is a "nice" old laptop, it still took 2 hours to do a genkernel. All I want it for is to run various networking utils on customer sites, like an mgetty over serial or tcpdump. What bugs me is the thought of lugging this thing somewhere, and finding out that I need to do a 6-hour emerge to get something I need really quickly. Perhaps this is a contrived example, but it shows my point perfectly. Why can't we get some sort of baseline install for generic machines. On this laptop, I don't care to have any optimizations. I was just trying to bring up the base install, use the included portage snapshot, and use all the packages from the GRP that I could. Anything else I could emerge. However, when I saw the 2-hour kernel build, I got scared off. The same thing holds true for my dual-Athlon file server. I don't care about optimizations. Even though it builds a kernel in 20 minutes, I don't want to manage a bunch of USE flags. I also don't want to get it all setup and running, and find that I need to take 3 days to compile the whole system because I found something I wanted that I didn't forsee. I like the customizations of Gentoo on my main machine, where I have the energy to hold the system's hand through all of the fine tuning. But on servers and utlity laptops, I don't care. My point is this: why isn't a "bigger GRP" a goal of the Gentoo project. If the idea of the GRP were expanded, we could get a lot more usuable system without having to compile anything. I think it would be awesome to take a snapshot of a standard-USE-flag, standard-optimization, i686 system, and produce a "distro" in the normal sense. That way a person can tweak out the system he wants, and run vanilla stuff on ones he doesn't care about. It would be the best of both worlds, and keep a person from needing to split their administrative skills between distros. PLEASE do not get me wrong. I'm not trying to start a flamewar. I'm just trying to figure out how I can make gentoo work for me on something other than a nice machine that I care to tweak out. I can see what the baseline compiler flags ought to be, but keeping up with the USE flags is wearing me out. I wouldn't know where to start to generate a list of the "big" ones that determine how a system really behaves overall. The biggest pain I see in this setup is the fact that I want security updates. In my own use, I was thinking about bringing up a vanilla system on my dual athlon, and building security updates with it to distribute as binaries to the laptop. I don't have any figures on this, but someone does. I'll bet 90% of the use of Gentoo is on i686 machines. Why wouldn't making such a snapshot or distro be appropriate? Thanks for listening, dk -- gentoo-dev@gentoo.org mailing list
