Hey! Welcome to my site. You may want to subscribe to my RSS feed. Thanks for visiting!
AJ Venter writes that we need more, not less Linux distributions. We’ve discussed it to a degree on the #clug IRC channel, where AJ Venter also sometimes hangs out as silentcoder. I wanted to discuss it with him, but he said that he doesn’t want to discuss it over IRC and also blogged that he doesn’t want to. He says that he really wants to discuss it over comments on the post instead. Since he has disabled comments on the blog post in question, I decided to write this blog-reply.
I’m not convinced that we need more distributions. More distributions would result in huge duplication of work:
- Additional bug trackers
- More packaging work
- Relationships between maintainers and upstream projects
- Documentation
- Additional installer work
- … and there’s probably a lot more
There’s also very little benefit from doing a whole new distribution from scratch. Doing a custom installation from an existing distribution has plenty of benefits:
- Existing installers
- Lots of existing packages
- Most common issues are known and can be tracked in the distribution’s bug tracker
Distributions such as Debian and Ubuntu are super-easy to adapt, and there are very few use-cases that could warrant doing a distribution from scratch as apposed to doing a custom install disc of those two systems. Ubuntu’s parent company, Canonical, even goes a step further by offering free hosting for free software packages via the Launchpad PPA service.
I don’t think AJ’s “diversity” “arguement” is solid or even makes a proper case for the need for more distributions. You can have diversity and satisfy a wide array of unique use cases by leveraging the work of the existing distributions, without being wasteful and duplicating effort unnecesarily.
December 14th, 2008 at 6:01 pm
As a recent convert I agree that more distributions isn’t so much the answer. I still don’t even understand why there is a KDE vs. GNOME debate, it all seems like duplication of effort to me. But maybe I’ll take some time next summer to figure it out.
December 14th, 2008 at 8:40 pm
We need look no further than Debian / Ubuntu to find evidence against creating Yet Another Linux Distribution. Ubuntu has relatively significant resources supporting it in the form of both Canonical employees and Ubuntu community members, and yet the quantity of working Debian packages that have broken versions in Ubuntu is ridiculously high. The packages in main have Canonical employees hovering over them keeping them in great shape, but the MOTU collective does a really bad job of keeping the “long tail” of packages in good working order. And Ubuntu isn’t even a “from-scratch” distribution.
Diversity is great, but reinventing the square wheel doesn’t give us diversity, it just gives us yet another useless wheel. The nature of free software is such that you do not have to reinvent the wheel to try out something new. Wasting your time badly duplicating other people’s work is an unnecessary distraction if you actually intend to try out new ideas: you’ll be swamped trying to deal with all the old ideas before you get anywhere.
December 14th, 2008 at 9:17 pm
Going back to the nature argument of which silentcoder is so fond, while we see great biological diversity, the way in which that great diversity is built involves appropriating large amounts of base structure from already-working constructs. So, you’re creating a new Linux distro — you’ll be using much of the same source code, much of the same root file structure, etc. So, why not go a little bit further and use the same package manager, same bug tracker, etc as a well-established distribution? Unless part of the innovation you’re trying to accomplish involves building a better bug tracker or better package manager, I think it serves the ecosystem better to reuse as much infrastructure (think proteins, DNA sequences, cell structure, etc) as possible while still adapting to fit a different niche.
December 14th, 2008 at 9:43 pm
It doesn’t really matter what we think since GNU/Linux is about freedom. We could debate this until we’re blue and all oxygen is inhaled, but it’s nevertheless a question of individual decisions. GNU/Linux evolve naturally; when visions and code by developers are appreciated by users it’s a success, if not a developer is still free to fulfil his own expectations. You can’t force a developer to do what he doesn’t want to do within a free developing model. Distributions with commercial backing are few, and won’t become many since economy is a limiting factor. Community distributions are many, but they do not belong to an organizational structure that can be controlled by some kind of “board of directors”.
To wish otherwise is equal to wish the death of GNU/Linux as we know it today. Without individual initiate on a community level it wouldn’t have existed. Without the choices many of the core Linux users of today wouldn’t have used it. The only way to unite the “brightest” developers to do “one” thing, is if the task by it’s own quality is so outstanding that there’s no doubt it couldn’t be done better. Besides that we would need to kill the human spirit and individualism before any surgery were done on GNU/Linux itself.
…
The diversity issue is exaggerated in my view. Besides some DE dependent applications, there’s few boarders hindering you from using the same applications in whatever distribution you choose. Some puritans wish to only use DE independent applications, while most don’t care if some KDE- or Gnome-libraries are installed or not.
December 14th, 2008 at 11:25 pm
I totally agree. The existence of hundreds of distributions scares windows users that are considering the switch to Linux and gives food to Trolls to spread FUD. Wouldn’t 100 distros be enough? Do we really need another 400(not well tested) being Ubuntu, Fedora or OpenSuse + codecs or X application??
December 14th, 2008 at 11:46 pm
I disagree… If you want to use the Debian base, then use one. But, for the sake of competition and innovation, the more the merrier. The more innovation, the better off Linux is. They might not set the world on fire, but some parts of a new from-scratch distro might make it into more mainstream distros. Why stifle this?
What would have become of Linux if Suse, or Mandrake never started up their distros because this same mentality was in place back in the 1990’s? We could have said that we had Slackware, Red Hat and Debian, so why bother with yet-another distro? The Linux community is better off because other ground-up distros did start up, and it’s the same thing now.
What exactly would they be duplicating? Even Ubuntu, which relies heavily on Debian, repackages everything in their repository. It’s not just Debian’s packages offered up to Ubuntu users. So, where is the savings of duplicated work?
Personally, choice is good. Therefore, I contend that we’re far better off with more ground-up distros than spinoff distros like Ubuntu. If I was stuck with just a couple of distros to choose from, and it wasn’t something I wanted, I’d go elsewhere. Luckily, I use a distro that fits me to a “T”, but isn’t one of the most popular distros out there. I couldn’t imagine if this type of mentality prevailed and my favorite distro didn’t exist at all…
December 15th, 2008 at 12:12 am
Guess it depends on what one is trying to diversify. If somebody is just going to do a prebuild of Ubuntu or Debian with a different mix of windows managers, well ok, your point is valid. But if someone is attempting to do a Debian instance but wants to use the Plan 9 file system rather extfs3 then what? It will be distinct and experimental and may yield an improvement not yet identified for the down stream distros.
Or what does one say of Linux Mint? Its a third level derivative (Debian->Ubuntu->LM). Yet because it has the ‘batteries included’ that Ubuntu decided to leave out for legal/political reasons LM has its adherents.
The fact is more distros do not have to lead to fragmentation. If it was agreed to in advance that Debian, Ubuntu and Linux Mint would all utilize a standard Main repository and deb branches for their unique requirements some of your concerns would be addressed. The core problem of the fragmentation is in the libraries as you mention. I don’t know if having more distros is necessarily the cause. There a plenty of packages lying around whimpering for lack of maintenance period.
December 15th, 2008 at 4:33 am
Less distros is better; please merge Ubuntu back into Debian and drop it. Well, the other way might be better since Debian is dying (possibly because of Ubuntu).
December 15th, 2008 at 9:34 am
[...] less, GNU/Linux distributions.Matt Fletcher on Why we need more, not less, GNU/Linux distributions.ssss on I am tired of IRC flamewarsjonathan carter » Blog Archive » “More Linux [...]