Bryan J. Smith wrote:
Oh no i wasn't agreeing or disagreeing.. I was just brainstorming and discussing the ideas in a general fashion and trying to give it some shape (for my own mind to)[ lots of snippage]
I think that was what I was saying.
But maybe I wasn't saying it as clearly?
My apologies ;-).
Eventually I'd like to see the "Quark" weed out a lot of stupid,Actually by splitting up the packages between CD1 = basic working server/workstation, and the rest on cd's 2..4, you've already filtered out most of those dependencies.
unnecesary dependencies in "Core." But that's a future ideal.
Secondly, i've tried this 'filter out silly dependencies' quite a few times my self.. It always ended up biting me in the behind though. (tried it on rh es, rh9 for some time actually). Often taking out the dependency will cause bugs to re-appear that were fixed by those dependencies, or it will break external 3rd party (or legacy) rpm's. It's easier to pick the low hanging fruit and just put all the 'silly deps' on the other cd's
I'd go the other way. No apps. Why? Because in short order, it willAh finaly something we disagree on :-) Small is suprisingly easy to accomplish.. However to make this install usefull to to the average user, thats something different i.m.o.
bloat to more than 1 CD. The idea here is to make it as small as
possible. Even if its only 300MB or smaller.
We might not share the same vision here.. I envision a 1 CD instalation that allows ppl to get up and running, and do all the basic server and/or workstation stuff that you would expect.. So basic IM, browsing, office document editing, graphic packages, full gnome/nautilus with basic utilities.. etc. Not the expanded list that makes it 4 cd's, but enough to have the same experiance as you would have using Windows 9x / XP.. even just a little more. For a lot of people this 1 CD instalation should already be enough to start 'using linux'. If after that they want to install another desktop envirioment, databases, etc.. then they will need the extra cd's or download things from internet repositories.
Thus my 'quark' (to use your definition) vision is different from yours.. While i like the concept of "just basics please", i can see this confusing people why they have to download the 4 cd's anyway (thus not solving that problem). While for advanced users this pure-basics 1 cd usefull, i think it misses the point for the "average user"
I gues a terminology i'd use to describe my 1st cd vision is 'assumed functionality'. An Average User(Tm) is the non-vocal majority who doesnt care to much about licencing politics, desktop wars or package inclusions.. All he wants is to download a CD, pop it in, reboot, and have a brand-sparkling-new linux system up and running within half and hour; Where he can do his buisness (mail, browse, open office documents, view his pictures, chat with friends, edit text files and play some media). If, after that experiance settles in, he wants video conferencing, databases or any of that 'extra' stuff, system-config-packages should point him the way to how to get that and ask for the instalation cd's or download it.. For ppl like that 'yum' is already quite a far fetched and scary concept
I gues the question i'm asking that appart from "having to be carefull not to bloat out of the 1 cd confinements for a basic install", what advantages would this "only very basics please" approach have over a "functional but kept to one cd" basic install concept?
Last argument i have on this is wasting CD space.. if CD1 only has 200Mb of stuff on it, there's a big risk this will add another cd to the instalation, and thus be more expensive to publish/burn, make for more cd swapping, and make ppl scream bloat even more then they do now..
BTW, does that list include Kerberos? I probably need to find out byOh no it didnt, add another 1 Mb to the number then (very minimal impact).. ps rpm-analyzer requires some hacking to get it running.. Using my tool works nicely to (ofcource i would think so), and shows very nicely what extra packages are pulled in to satisfy dependencies.. great for if your doing a measuring game
using RPM-Analyzer.
I see little reason to include development. They can be fetched via Apt/Yam.Self building and able-to-compile systems are very open-source and *nix like though.. yum/apt is great, but really relying on them to much is not a good thing either.. Using yum to install mysql .. sure... Using yum to make assumed basics functionality .. goes to far for me.
For those that can't, Quark is not for them. They should install the full
"Core" instead (possibly from 3rd party DVD with Extras added).
Reality is that a lot of users, experianced or not, like compiling things from source, or are being told to do so in emails, faq's and todo's.. It's also one of those 'mystical' new linux user experiance when they see they can.
Not to mention using yum to forfill building dependencies is a nightmare for anyone who's not a heavely experianced redhat user.. What -devel package gives me /usr/include/obscure/file.h or /usr/lib/weird/name.so ? Yum doesnt process build requirements.. So you'd have to teach them about --whatprovides and --redhat-provides type command lines or teach them what packages provide what libs.. Neither of those options is very attractive if you ask me.
The only reason i see to not include devel packages in an instalation is for either a secure server (if you can't compile your prefered root kit, 90% of your h4x0rs are already stomped) or for corperate desktop style instalations.
Actually in an enterprise environment i'd use the anaconda-ks.cfg kickstart file, adjust that to my needs as i see fit and make a new install CD from it (the hard way), or just network instalation using a bootdisk or preferably thru PXE (the easy way) and install it that way.In an enterprise environment, I'd rather have a "base" install like Quark on a CD, and then run a script that does an "apt-get" of everything else needed. But that's just me.
