We will minimize API changes along the way, which will allow us to
change everything behind the API, and do new an shiny things.
I hope that makes sense.
Chris
+1
> We will minimize API changes along the way, which will allow us to
> change everything behind the API, and do new an shiny things.
I think that's the best way forward. We're packing in so much
flexibility from the start that the API also has to be well-defined to
handle change.
Owen
It's a balance in the end, and the only question we can decide at this
point is which side of the spectrum we'll default to. Like all fresh
shiny-faced start-from-scratch types we're probably best off deciding
that APIs are contracts and should be phased out slowly after being
clearly deprecated and any undocumented behavior is up for grabs in
terms of how something will function in the future.
The reason this is such a problem is that 'current' plugin/theme
authors are A-list/high-priority users of the software, just as future
extenders and end users are. Which way do you sway? MSFT's business
model for windows depends on "developers, developers, developers" so
they sway towards being conservative.
We really need to step back and reinvent the whole plugin/theme
ecosystem management approach so that it's easy to spot and report
plugins that aren't working with a release candidate, help the
developer fix it, etc.
Actually that's an idea isn't it? Backwards-compat breakage is usually
just about functionality that can be recoded without much
decision-making/fuss. If there's a culture of making sure all 'listed'
plugins/themes are tested and someone does the quick fix if the
maintaining developer doesn't currently have time, and so on, this
ends up being a social problem mostly.
--
Firas Durri | http://firasd.org