On 10-09-08 1:49 , Joshua A. Andler wrote:
Yes, we do want people to use new features. I myself want to use them. :) I think that if we have things implemented in a sane way to get non-finalized features in (especially taking into account that there is no proposed syntax for a lot of these things yet even), that we can possibly help to ensure that the features DO get added to the spec.
My opinion is based on the User experience side. Please excuse my ignorance of the possible issues it might/might not introduce in the programming field. I might be completely off here. :) I think that the implementation is not too hard from the User Experience point of view. Additionally, it could allow us to gather some insights from the users that we can use to improve UI, too. A simple example, presuming that there's no 'experimental' branch, but one Inkscape, that allows the widest user-base, would be:
1. Not to let down the trust and loyalty of the people, a stable Inkscape release should be all nice and by-the-book, out of the box, I think.
2. The new features are introduced with a honest and clear, but serious approach. E.g. User enables the feature(s) explicitly using Inkscape preferences' 'Experimental' panel. The panel states: -what is the feature the user is turning on -info on the implications, and the fallback -a link to more info online (it would be great to allow user to turn off the feature, too :)) The panel might have a prominent black/yellow stripe pattern to indicate the 'experimentallity' of this. (or some color...)
This explicit acceptance gives the user a clear signal and choice of crossing from the 'safe and stable' into the 'experimental area'. It's like opening the hood of the car, removing the back cover of the TV set, or lifting a lid on the pot to add some more pepper to mother-in-law's soup. ;)
3. Any new UI element that might appear following this user's decision is visually labeled to clearly indicate the 'experimental' status. For example, using the same pattern or color the user sees on the 'experimental' panel would provide a clear correlation.
4. The user should never lose the data, so the fallback Krzysztof suggested, plus a fair notice when saving or opening such a file, would be nice. A link to a shoutbox, of even an ability to send feedback directly from Inkscape, would be awesome.
From my experience, an approach like this indicates the user a clear 'crossing line' and removes (I'd say) all the confusion and helplessness. Additionally, with an ability to send feedback, it turns user's frustration into collaboration. SVG people could and try and see the thing too and maybe speed up some things! ;)
... The ability to test features live, real users using the real stuff (and by 'testing' I mean their code, implementations and acceptance) is a huge advantage the open source has over any other development paradigm. Excepting evolution ;). We should embrace this more, I think.
The issue of standards is and always will be a love/hate thing. But, as Joshua said, Inkscape is a factor in shaping of these standards.