PS: I personally feel that by the time a package author presents a well-written
and fairly complete package for inclusion into ELPA, it is too late to make the
argument that they should have instead contributed that functionality to a
built-in package. If you can convince them earlier in the process, of course,
but it certainly adds to the outside impression that ELPA is a gauntlet that
must be run, which I think all agree is not desired.
I’m not sure I follow. Most of the time package authors don’t ssy anything
until they want to publish the package, so how can the maintainers suggest to
them to contribute to an existing package?
An alternative would be for potential package authors to post to emacs-devel,
stating that they would like to see feature foo in package bar. If it turns
out that it is not feasible to implement the feature, then sure, a new package
can be created.