wattOS R8 Is Now Based On Debian Rather Than Ubuntu

[chrisb]

So from what I read here for you a distribution comes down to: ...

I don't have a strict definition, I am adaptable.. It might make more sense if the world used a stricter definition based on who provides the packages, but the world does not : http://en.wikipedia.org/wiki/List_of..._distributions

which by the way, are, with exception of Kubuntu, not supported for 5 years, read the Xubuntu release notes

Respins with 5 year support: Ubuntu Desktop, Ubuntu Server, Ubuntu Core, Kubuntu, Edubuntu, and Ubuntu Kylin

The other respins will still be getting hardware enablement and backports for 5 years. The thing they will not get is community security fixes for packages in the universe repository after 3 years. Official security updates for packages in the main and restricted repositories will be supported until the EOL of the 14.04 repositories, regardless of what respin is in use.

[Vim_User]

Sure, you could start with Debian Testing or Debian Sid, but those are both unsupported and not recommended by the Security Team (Debian security FAQ: "If you want to have a secure (and stable) server you are strongly encouraged to stay with stable.") Whereas if you start with Ubuntu LTS, you will get security updates for 5 years. And if you base off Debian Testing or Sid you will have to periodically sync packages to get hardware enablement and bug fixes.



You are drawing a distinction between a "respin" and "distribution", where a lot of people do not make such a distinction. Distrowatch lists what you call "respins" as distributions eg. Xubuntu "Xubuntu is a Linux distribution based on Ubuntu." Anyway, some users see a lot of value in distributions/respins like Xubuntu and Lubuntu.



If you base off Debian Stable then you have to do hardware enablement yourself, because Debian do not do hardware enablement for stable releases.

If you base off Debian Testing or Unstable, then you have to do security tracking and fixes yourself, because the Debian Security Team do not handle Testing or Unstable.

If you base off Ubuntu LTS, then you will get 5 years of hardware enablement and security fixes from Ubuntu.



"Well tested" is subjective. What Ubuntu does have is several million desktop users globally. Testing is not just about the internal testing of Canonical or Ubuntu, but about the fact that there are millions of users out there running the same software stack. Sure, it's not perfect, but those millions of users do catch an awful lot of bugs. (Bugs which are now automatically uploaded and backtraced by errors.ubuntu.com; Debian could really benefit from having something similar.)

So from what I read here for you a distribution comes down to: Create a meta-package that pulls in all the packages we want on a minimal install of an already existing distro. Or just install them yourself and create a medium with installer using Debian Live/Remastersys/whatever. This would make actually make anyone who installs a package that is not in the default set of the base distro a distro maintainer. I wouldn't go so far, but if that is how you define distro, then so be it, I won't. That is by the way exactly what Xubuntu/Lubuntu/Kubuntu/... do, only that their meta-packages (like the xubuntu-desktop package) are integrated in the base distro's repositories, making it nothing more than a respin (which by the way, are, with exception of Kubuntu, not supported for 5 years, read the Xubuntu release notes). This makes it indeed easier to install such a respin, but there is nothing in Ubuntu that hinders you to get to the exactly same point without that meta-package. This means there is indeed no real value added. Don't get me wrong, I don't want to play down the work of the people working on those respins, but they are IMHO nothing more than that. Fedora, for example, does it right and calls their respins respins, not distributions.

We can now discuss on and on on what makes a distribution, but I doubt that that makes any sense, you have your definition and I doubt that it will be changed by telling you my definition.

[ext73]

I do not think that They have achieved better results

http://www.netext73.pl/2014/04/kernele-v29x-juz-sa.html

http://www.netext73.pl/2014/03/kolejny-may-rektord.html

[chrisb]

Or you could just start with Debian Testing or Debian Sid, as Ubuntu and many other distros do.

Sure, you could start with Debian Testing or Debian Sid, but those are both unsupported and not recommended by the Security Team (Debian security FAQ: "If you want to have a secure (and stable) server you are strongly encouraged to stay with stable.") Whereas if you start with Ubuntu LTS, you will get security updates for 5 years. And if you base off Debian Testing or Sid you will have to periodically sync packages to get hardware enablement and bug fixes.

Still, what you describe is how to make it easy to do respins, not to create an actual distro. What would be the advantage of a distro that only has the packages provided by its base system? None.

You are drawing a distinction between a "respin" and "distribution", where a lot of people do not make such a distinction. Distrowatch lists what you call "respins" as distributions eg. Xubuntu "Xubuntu is a Linux distribution based on Ubuntu." Anyway, some users see a lot of value in distributions/respins like Xubuntu and Lubuntu.

So in short: For providing a respin Ubuntu may be the better base system (though I would not call any version of Ubuntu "well tested"), but for creating a new distro I don't see that it is a better base than Debian or many other distros often used as base.

If you base off Debian Stable then you have to do hardware enablement yourself, because Debian do not do hardware enablement for stable releases.

If you base off Debian Testing or Unstable, then you have to do security tracking and fixes yourself, because the Debian Security Team do not handle Testing or Unstable.

If you base off Ubuntu LTS, then you will get 5 years of hardware enablement and security fixes from Ubuntu.

(though I would not call any version of Ubuntu "well tested")

"Well tested" is subjective. What Ubuntu does have is several million desktop users globally. Testing is not just about the internal testing of Canonical or Ubuntu, but about the fact that there are millions of users out there running the same software stack. Sure, it's not perfect, but those millions of users do catch an awful lot of bugs. (Bugs which are now automatically uploaded and backtraced by errors.ubuntu.com; Debian could really benefit from having something similar.)

[stiiixy]

I've been forced to test my own advice over the last few days, as my desktop's PSU died.

Have a Dell Tablet with 8GB RAM, and a 11W TDP i5 Chip (gaming seems to max it at 8W as the cooling isn't good enough). Runs a VM fine if I need it, and surprisingly does all I want except play games (That being a pro in my books).

I hear ya. I haven't looked in to tablets recently for general desktop computing, but these dyas most people would definately do something like that if it could swing both ways nicely. Are there any low-cost low power products out there I could investigate?

[Vim_User]

Or you could just start with Debian Testing or Debian Sid, as Ubuntu and many other distros do. Still, what you describe is how to make it easy to do respins, not to create an actual distro. What would be the advantage of a distro that only has the packages provided by its base system? None.
I expect from someone creating a new distro to add value to the base system besides having a different mix of packages installed. For example, like Kali, based on Debian, but with custom compiled kernel and software. Or like Mint, adding their tools to the Ubuntu (and Debian) base. Or like Salix, adding dependency based package management to Slackware, ... .

So in short: For providing a respin Ubuntu may be the better base system (though I would not call any version of Ubuntu "well tested"), but for creating a new distro I don't see that it is a better base than Debian or many other distros often used as base.

[chrisb]

You could use the backports repository to get a newer kernel, but what I see here is nothing that would make creating a distro based on Debian harder than on Ubuntu. What you describe is that is seems indeed to be easier to create a respin (read: a different mix of packages with no real additional value) and as it seems wattOS seems to be exactly that, Debian with backports enabled and some packages pulled from Jessie (which IMHO is by the way a bad idea, they should rather backport).

Debian backports is not the same thing as Ubuntu LTS updates. backports has to be manually enabled by users, so the packages there are not as well tested, eg. the packages there are not used for the livecd, are not used for new installs. In particular, there is no backport of Xorg, where hardware enablement is very necessary. The enablement stack in Ubuntu is pushed to -updates, so it gets installed by everyone by default, and is hence much better tested. 2-year-old Ubuntu LTS has better hardware enablement than 1-year-old Debian Stable:

Debian Stable, originally released May 2013, xserver-xorg-video-intel=2:2.19.0-6

Ubuntu 12.04 LTS, originally released April 2012, xserver-xorg-video-intel =2:2.99.904

Sure, you could start with Debian Stable, and backport a new kernel, and backport Xorg, and Xorg drivers, and then fix everything that you broke by doing that, and then you could start to backport any other libraries or applications that you need which are not recent enough in Debian Stable... and then after release you need to keep track of important bugfixes and security updates for all of your backports. And after all that, your users will be running a unique combination of packages which will probably have some compatibility problems. Or you could use Ubuntu, and let them worry about doing all of this hardware enablement stuff, and your users will be running exactly the same combination of packages as every other *buntu user.

As I said, Debian is a great project, but the fact that the stable release won't even boot on a 1.5 year old laptop due to lack of hardware enablement is a huge problem for normal users, which in turn is a problem for respins.

Btw, someone did file a bug about hardware enablement for Intel Haswell in Wheezy: bug #714203 xserver-xorg-video-intel: No Haswell support in 2.19.0. The reporter mentioned that the package in experimental repository worked and got the response from maintainer - "So if it's already working I'm not sure what the point of this bug is. Closing." Who cares about hardware enablement for stable releases when there's an experimental package that is not used in the stable installer, or stable installs? Every user should just know that they have to roll their own installer and incorporate experimental packages if they want to install Debian stable on a new laptop, right?

[ncopa]

Alpine Linux is the most energy-efficient distribution. You can benchmark it and compare with others.
http://alpinelinux.org/

Can you back it with any numbers? Has anyone actually tested this?

[zeealpal]

I usually recommended people buy themselves a smaller low end laptop for this very reason. Plus, they get portabilty AND battery-back-up, too boot. It's still fairly relevant now,. considering the options out there (even more so with a business moel that can support the dsktop kits). Support at times can be a bit of a bummer, and a little research goes far.

But still, nothing like buying a monitor with this micro-gear built in already. Or a NUC-style with a VESA mount to throw on the back of said unit. I wouldn't mind a VESA mount with double-sided options to then mount it on walls as well, but that's something I cna knock up easy enough I suppose with a sheet of aluminium and a drill.

To many options these days!

I've been forced to test my own advice over the last few days, as my desktop's PSU died.

Have a Dell Tablet with 8GB RAM, and a 11W TDP i5 Chip (gaming seems to max it at 8W as the cooling isn't good enough). Runs a VM fine if I need it, and surprisingly does all I want except play games (That being a pro in my books).

[stiiixy]

If you really wanted to save power, just run one of the Intel Nucs with < 10W TDP, or the new AMD APU's with an SSD, presuming that provides enough power to do the tasks you want. Then get an energy efficient monitor (smaller uses less power as well) and a wired non backlit KB/Mouse.

There is far more power to be saved by selecting specific hardware rather than specific.

Any software differences can often be configured manually, and disappear when the system is under load.

I usually recommended people buy themselves a smaller low end laptop for this very reason. Plus, they get portabilty AND battery-back-up, too boot. It's still fairly relevant now,. considering the options out there (even more so with a business moel that can support the dsktop kits). Support at times can be a bit of a bummer, and a little research goes far.

But still, nothing like buying a monitor with this micro-gear built in already. Or a NUC-style with a VESA mount to throw on the back of said unit. I wouldn't mind a VESA mount with double-sided options to then mount it on walls as well, but that's something I cna knock up easy enough I suppose with a sheet of aluminium and a drill.

To many options these days!