Am Mittwoch, den 24.02.2010, 23:38 -0600 schrieb Matt Domsch:
These questions were originally posted to the Spins SIG a couple weeks ago. At the following SIG meeting, members suggested that these questions would be better posed to the individual spin owners, or teams that are working on the spins, in addition to the SIG.
The Spins SIG consists of the spin owners and we already discussed this in the Spins SIG meetings of the last two weeks. I think most of the spin owners already outlined their point of view in the meetings, some also relied to your mail.
In addition to this I already wrote something on the board's mailing list last week. Nevertheless I will try to answer the question once again, this time a little more detailed.
[snipped]
Board-level Question: Can Spins/SIGS or Fedora remixes define their own target audience?
Background
[snipped]
Possible Solutions
- Board sets target audience broadly, spins tailor to a subset
thereof.
- Board sets target audience broadly, spins tailor to either a
subset thereof, or to additional audiences outside that scope so long as there are no conflicts.
- Board does not set target audience, leaves it to each Spin to
set their specific target audience. 1. requires spins to be much more than consumers of Fedora content. 2. audience of some spins may overlap. That's OK. 3. audience of some spins may technically conflict. How to resolve conflicts? Spins -> FESCo -> Board.
As members contributing to Spins, how do you view the above, and how would you like to see Spins interact with the larger Project with respect to defining target audiences?
I don't really understand all this discussion about a target audience or potential problems. The definitions are on a completely different level.
The board is the political body of the Fedora Project. Thus their definition of a target audience is a political decision. The Spins SIG on the other hand works on technical implementations, this means their definition of target audience is more technical.
So far, the board has defined a "working" target audience. Early adopters who always want the latest and greatest technology and who are not afraid of changes. Active users who are interested in development. People who don't mind searching the net for a solution to a technical problems. People who like to contribute.
I think all spin maintainers can subscribe to this point of view, otherwise we would not contribute to Fedora. Same goes for the users: They may want to use this or that spin, but first off all they want to use Fedora as such. If they don't like Fedora, they will not use our KDE Spin but another KDE-based distro.
Given the boards definition remains political, I have no problems with ether one of the possible answers. The technical implementation of the spins is automatically a subset of the board's definition. But if the boards starts defining technical goals such as "lets prefer GNOME users because Red Hat contributes to GNOME a lot and the Desktop Spin is our flagship product" we will run into trouble. Then the spins must be able to define their own target audience, even outside the audience defined by the board.
Let's face it: Although we already have a target audience, we have case 3.3: Spins technically do conflict. I will explain this below. What I want to say is that IMO the answers lack another possibility: The board sets a target audience but still there are conflicts.
So before I can decide on one of the possible solutions, I'd like to hear a definition of "target audience". I know this is a bit of a chicken and egg problem, but I think at the current state the board should be able to at least say on what level will this audience be defined. Political and theoretical or technical and applied?
Board-level Question: Can Spins/SIGS or Fedora remixes change the code enough to meet their goals?
Background
[snipped]
Given the present situation:
- Has any Spin found the present situation unduly restrictive?
- If so, how specifically?
A spin is defined as installable Live-CD. This means it must ship anaconda and firstboot. firstboot requires system-config-keyboard requires metacity requires GConf2 and tons of other GNOME stuff. There are other dependency chains as well (e.g. notification-daemon), but this is the worst one.
- Has any Spin found they cannot address their target audience properly?
- If so, in what way?
Xfce users want Xfce, LXDE users want LXDE. Both want a lightweight system, otherwise they would likely be using GNOME. But what they get is more or less GNOME. Without all the GNOME packages we would have significantly more space on the media.
2. Is the root problem that all packages must be in the official repositories?
No, the root of the problems are the GNOME maintainers refuse to consider other desktops and their users.
I'm convinced that we can address these issues on a technical level in one repository. OpenSUSE already does a great job here: They have different branding packages wich contain the SUSE specific changes. This is similar to what we do with fedora-logos, but it's a more complete approach.
For example, there is a package called gconf2-branding-openSUSE which contains all the modified GConf schemas. By replacing this package, you can change the complete settings of the GNOME desktop. This would be useful for us too, think of a GNOME based Fedora Mini Spin for Netbooks, that wants to use another other panel layout. We cannot do this in Fedora ATM, but it is possible. For more info about OpenSUSE's branding, see http://en.opensuse.org/Packaging/Branding
We must not break up into different projects like Ubuntu with Kubuntu, Xubuntu and Lubuntu because this results in a duplication of work and incompatible packages. Instead everything should happen under the umbrella of the Fedora Project. We need to work out the problems on a technical level instead of forking. I'm sure we can do it if all desktop SIGs cooperate.
- How are you addressing this today?
We cannot. Currently *all* spins must ship metacity and all it's dependencies.
- How would you like to address this in the future?
With more fine grained packaging. Introduce a virtual provides for window-managers instead of hardcoding one. system-config-keyboard maintainer refused to do this in the past, but now that we have the critical paths set for each desktop, I don't see why the virtual provides would introduce problems.
- Are the resources you would need readily available?
- If not, what would you need to properly address this?
* Webspace for direct downloads of the spins. not only torrents. * A way to count these downloads. * Officially labeled live media of the spins for promotion. In EMEA we currently have GNOME, not even KDE.
- Is the transition from "Spin" to "Remix" onerous?
- If so, what can be done to make it less so?
I don't think the transition from a spin to a remix is onerous, but from a remix to a spin it's problematic because one has to follow all the Fedora guidelines and restrictions. This is why I think we need the remixes and we cannot really forbid them.
However I think we should not advocate remixes, at least not if they are questionable. I never understood why the board approved the Fedora Russia Remix and allowed it to use the Fedora trademark. It contains non-free software or codecs and thus violates our very basic foundations.
As long as a spin follows the Fedora foundations, I have no problems with it. Maybe it's necessary because some things are technically not possible in Fedora ATM. This is one more reason why I think we need to work on solving technical issues with the spins in Fedora: The more free the spin maintainers are, the less reason they have to do a remix instead of a proper spin.
Regards, Christoph