[kde-community] RFC: Distribution outreach program

Philipp Stefan sogatori.ml at gmail.com
Sun Jan 31 15:49:36 UTC 2016


Hi Thomas,
I must say I am sceptical of the idea.
> Hi everyone,
> distributions are - and most likely will be for the foreseeable future - be
> the main channel by which users get our software. Therefore, we're eternally
> grateful for the job they are doing.
> However, it happens - too often for my taste - that users face problems while
> using our software which only happen in their distribution. We could say
> "well, that's not our problem, then", but of course our goal is to provide our
> users the best possible experience, so blaming others for problems does not
> help us.
I don't see how this is "blaming" someone. If a bug is not in your software then there's only so much you can do about it.
On the other hand there are also plenty of "upstream" bugs on the distributions' bug trackers.

>   I believe that we can improve the situation by intensifying our cooperation
> with distributions even further. On the other hand, however, distributions
> also have to do their part in order to make our software run optimally on
> them. Of course we're not really in a position to demand anything from them
> per se: If distributions make decisions which screw up our software, we cannot
> effectively keep them from doing so.
>
> What we _can_ do, however, is offer them incentives for doing so. This leads
> to my idea: Establishing a "Distribution Outreach Program", coupled with an
> official "badge of approval". How could this work?
> 1. We define - as strictly as necessary - criteria which a distribution has to
> meet in order for our software to run optimally on them. These criteria could
> include things like
>   - How fast do they have to deliver our newest major as well as minor releases
>   - Which version of our dependencies they should ship with each version (not
> just the minimum dependency, but the version we've identified to work best)
>   - With which options they should compile and package our software
>   - Other measures to avoid common downstream problems we've identified
> 2. Distributions can agree to comply with these criteria
> 3. They officially become part of the outreach program which gives them
>   - More direct help with problems that may still arise
>   - A badge like "Works great with Plasma" or "Works great with KDE
> Applications" which they can put on their website
>   - A position in a special section onhttps://community.kde.org/Distributions  
> 4. We check if they really meet the criteria and if they stop doing so, there
> is a defined procedure for notifying them and eventually kicking them out if
> they continue to fail complying
The problem I see with this are the following:
	1. There will be no catch all criteria that won't exclude or at least penalize some distros because of how they are set up.
	(Think release cycles that do not sync up, unrelated blocking bugs etc.)
	2. Many distros do not ship KDE software only. A situation might very well arise that a distro has to compromise to solve a
	situation at a certain time. It will force distributions to either stick to our criteria or lose that juicy badge of approval.
	3. The bureaucratic overhead. The distros have to be tested regularly, there has to be some kind of official process to appeal
	a decision that distros feel are unfair and the badges have to be updated too. Knowing how well we are doing with updating
	as vital things as documentation it doesn't give me positive vibes to add another project to the list.
	4. There's no real penalty. If you feel the need to give out badges that say "Our software runs well here" than you should
	not be afraid to say "It doesn't run well on distro Y" not just implicitly. When there's no penalty for not complying and
	no incentive other than appearing on some web page most people will probably not even look at before we update our websites,
	then there's no reason to join other than good will.

I believe that it would be much better to just have a guide a la "Common mistakes made when packaging KDE software". It would
not define any criteria that will lead to public endorsement or disapproval and it will also spare us all the paperwork.

Cheers

Phil




More information about the kde-community mailing list