On Thu, 2005-01-27 at 23:23 -0700, Tyler Larson wrote:
Rodd Clarkson wrote:
Lets face it, if we're really interested in shrinking the size of the distro, cutting a few packages like XMMS isn't going to make a spit of difference. Not even eliminating unused packages is going to do much good.
Sure xmms is small, but it's about 4.8M of RPMS (including devel, flac and skins) On top of this is requires the GTK-1.x libraries so assuming not a lot else uses these libraries you're another step towards not needing them in core anymore.
And while 4.8M doesn't sound like much, that's almost 1% of a CD-ROM. It's a start. You only need to find about 120 more little apps that are duplicated in functionality and your well on your way to a CD-ROM being removed.
If we want to make an appreciable difference in the size of FC, we've got to cut at least enough to remove a CD or two. If we just cut a few redundant packages, we piss some people off, but gain nearly nothing.
Agreed. But as long as these packages are easy to install (yum install package) then this isn't a huge deal.
If we really want to cut the size of the distro by more than a few hundred meg, we've got to start removing functionality. Plain and simple. Something that lots of people *need* has to go. And you're not going to do it without making a whole lot of people really, really mad.
Not if you handle it well. I know that's a big ask, but it can be handled well. There a plenty of distros out there that only use a single CD and these users are more than satisfied with what comes on the disk.
So, what gets the ax? Tough to say. There's only 15 packages in fc3 that are >= 20 Meg. And OOo accounts for almost 200M of that. But can we really ax OpenOffice? Heresy. KDE? Blasphemous.
Sure, stick OOo on a extras CD-ROM. The ISO would be small, so those who want it wouldn't have to download a lot, and most would just yum install it. Since RH really doesn't focus on KDE (all the RH tools use GTK) I don't have any problem with KDE being on a separate Extras disk either.
This may sound arrogant, I'll admit that I'm not a KDE user, but haing to grab a single extras disk for KDE (and a lot of KDE users sound like they get their KDE from somewhere else than RH) instead of everyone having to grab four disks is a big win for everyone, even if you want KDE. KDE users grab two ISOs, others grab one. Users of OOo grab another. Sure you might end up with 9 or 10 extras disks (Java, KDE, OOo, etc) but none would fill a disk so you aren't grabbing huge amounts of stuff your aren't ever going to use.
Imagine how nice is would be if instead of trying to fit everything onto four disks inside four ISOs with a stack of software you are never going to use so you can get the few bits you do use, you could instead grab on CORE ISO and four or five smaller isos that target the stuff you want.
Sure, we could move packages to Extras... if only we had some idea what Extras is. Whatever it is, though, I'm quite certain that the packages *I* use don't belong there, right? But if the Extras CDs are distributed with the core CDs, and if most people will need the Extras CDs to get those last three packages they're rather fond of, what really is the point dividing them up? Other than, of course, to inconvenience the user. I was all for it a few hours ago, but now it doesn't sound like we'll be helping anyone out.
If people need a few extra packages (like the three number you suggest) it's going to be far easier to just yum install them, rather than download three extra 650MB ISOs for these three, or four, or ten, or twenty packages.
The way I see it, we're left with two options: A) Big distro. Deal with it.
Little distro, better deal
B) Piss lots of people off because Fedora no longer includes the software they use. Sorry KDE and OOo users.
Well defined ISOs which target users needs and save dramatically on the amount of downloads they need just to install CORE, KDE and OOo. Everyone wins.
The third option, "remove the stuff people don't use", seems more like a pipe dream than a viable course of action. You can't remove enough fat to keep A from happening without bringing about B. Moving a substantial amount of stuff into an "Extras" category (or whatever it is) seems like a "worst of both worlds" solution. Not only has the total software base not shrunk, but it's now more difficult to find the package you want. What are the chances, really, that the average user *won't* want at least some of the software on the Extras CD(s)?
Everyone uses something different so it just wont happen. Personally I would love to just have to download the CORE disk, and the OOo, Java, and Devel disks and then just yum the few other packages I need. What a dream. Having to download over 2GB of disks is a PITA.
Of course, the Extras concept has to work well and people have to realize that there is value in having ISOs that don't consume a CD. If all you need for OOo is a 200 ISO, then that's all it should be. The temptation to do OOo and KDE on the one disk should be ignored. Both ISOs won't be much more than a single ISO with both, but users who only want one or the other will be much better off.
I don't know. The more I think about it, the less excited I get. Perhaps someone can share a more rosy picture of how moving packages to Fedora Extras is supposed to make the users' lives so much easier. Anyone care to bite?
The more I think about it the more I think this is the way to go. I'd love a distro that offered more than a single disks functionality, but also offered a single disk that functions, along with small(er) downloads to address particular focus areas. Combine this with the ease of using yum to add those one or two other packages and how good would that be.
Rodd