User talk:Rabauke

From openSUSE

Thoughts on openSUSE's focus

Nobody will disagree that focusing is a good thing, yet foucssing on being "the best distro" or "archive high quality releases" is not really focusing, because a focus, by definition, is limited to a certain space. However, if you have enough resources, you might have more than one focus, yet not at the same time.

If you want to be the best, you have to focus on one objective at a time, set-up everything within the scene and take the picture. You will get a sharp and detailed picture of that objective. After that you can move on to the next objective. The ground that the objectives base on might still be the same, yet at another place and time.

If you do not stick to the above your picture will contain both objectives, yet either one of them is really blury and the other sharp, or both are blury.

To illustrate this: Objectives are different types of trees on a field. Users of that objective sit in the trees. The ground is the base-system.

If you take a picture, you might take a picture of all objectives, landscape, which will force you to increase the distance, thus all objectives, including their users, become less detailed and smaller, you might not hear what they say or see what they do, you might not even recognise them, neither do they know who you are or how to communicate with you. Depending on their distance (releases) to the picture-taking person, they might have a different degree of sharpness.

If you'd just take a picture of one tree at a time, you have a more detailed and sharp view of the tree and its ground, including the users sitting in it. You might even have several photographers taking pictures, yet those will still be sepearte pictures at a different point in time ans space.

So you have to make a choice, either be always less sharp and detailed than others and hence never be the best for a certain objective or focus on one at a time. There is a third option, which is out of reach for a project like opensuse, that is, provide as much resources as necessary to get a huge picture with perfect depth in field.

If openSUSE wants to be a does-it-all distro, it will fail, because other distros will be better in every area but the "does-it-all", which hardly anybody uses. Who runs Gnome in parallel with KDE and on top of that a web- and email-server, i.e. does it all? Even for trying out different DEs, the openSUSE DVD is not the best distro, because there are live-CDs.

So here is my thought:

Server focus

This is a slow moving target and does not need a new release every six/eight months and it is independent of current DE releases, graphics/wlan/webcam-drivers. It does not care whether the current KDE release does not work yet or whether pulseaudio is still immature etc. Of course it should be possible to install a DE on this release, yet it does not need to be polished and integrated, so it should just be the most stable release of that DE, no matter when/whether that release was part of an openSUSE release. A server does not need integration with the DE.

So the difference compared to the SLE releases is that a server focus does not care about the DE. Currently SLE includes the DE releases of a certain openSUSE release, even though that one might not be the most stable release of that DE. Developers need to backport fixes, instead of just using/building the upstream version that already includes those fixes. And for what? Because of the openSUSE integration/branding of that DE -- on a server.

Desktop/Notebook focus

This is a quick moving target and as such not coherent, because DEs are released at different times. Hence this needs to be split. Most (and that's what one should focus on) desktop/notebook users do not need any server software, neither do they need more than one DE or seamless mix and match of every tool there is. What they do need is a recent kernel for their new hardware and a recent high quality release of their DE, that's it. If the high quality upstream release of KDE happens two months after the one for Gnome, so be it. Pushing them into one will only cause KDE users to replace the shipped KDE version with the one from buildservice anyway, so why spend time on backporting features/fixes from a not-yet released version to have both DEs released in the same openSUSE release?

As mentioned before, different DEs have different release schedules, so forcing them into one release cylcle is just not going to work. Thus there needs to be a split.

KDE focus

Release a CD that only includes KDE and no Gnome applications. It does include common gtk apps like Gimp, but not Evolution, Mono, beagle and alike. It does not include any windowmanager but KDE's. People that get this release will know that it is a KDE release and not a server or mix-and-match release. They can install whatever they want from the buildservice, but that's not what openSUSE has to worry about.

Taking susestudio into consideration this would mean that the kde-team releases and recommends a KDE CD for KDE users, not caring that much about the DVD release. for openSUSE 11.1 this would have meant to just include KDE 4.1.3 without any work on backporting fixes and features, but just plain KDE 4.1.3 packages. Instead of wasting the time on backporting, the team could have put all their resources into polishing KDE 4.2 and releasing a KDE CD in February using the packages from the buildservice.

This would not be maintenance hell, because bugfixing for KDE 4.1 did go on upstream and 4.1.4 packages could just be supplied via YOU. So regarding a DE openSUSe would become a rolling distro, supplying the latest bugfix-releases for the DE version shipped, which just involves packaging and not backporting. The only thing that would lack, might be branding.

The liveCD would include KDE 4.2 and get KDE updates from the buildservice (a different repo than the one the devs are working on), after every bugfix release of KDE 4.2, so no backporting needed here either.

With KDe 4.3 there would be another liveCD, with a new base-system in case a new openSUSE version was released in the meantime, which would then include the KDE 4.2 packages previously used on the liveCD.

Gnome focus

Release a CD that only includes Gnome and no KDE apps. Same reasons as outlined above. People that want to use Gnome, want a high quality Gnome release and if that is not possible at the time of a high quality KDE release, so be it.

Drop everythign else to the buildservice

Offer a GUI to search and install from buildservice, as download.opensuse.org/search currently does, but more refined. So if I installed the KDE release and want ekiga, I go to that GUI and search for ekiga to install it and add the repo, without having to know which repo that is. This GUI could be integrated into YaST.

Drop xfce and alike from the releases and only offer them via buildservice. the community will offer installable Live-CDs of those if needed.

Drop games that are not part of a DE.