I suggest we do not try to reinvent the wheel.
Simply take a look how existing, large communities do this kind
of work.
Example: https://jcp.org/en/jsr/all
For each large feature (a good example in the D case would be
pattern matching, allocators, reference counting, additional
garbage collectors, etc) a special working group should be
formed. With initial working group owner who manages the group
(adds other owners, collaborators, or some other types of users).
Each group would work at own pace until they come up with a good
design, and after that the group does the implementation as well.
Each group would have own repository where they commit their
work. Organising such groups can be uniform (something I think is
better), or each group organises itself.
Naturally, it is probably a full-time job for someone to
coordinate with groups, does overall monitoring, etc.
Each group can, and probably will, gain interest of the academic
world, something we should encourage, and help PhD/MSc students
jump in relevant groups to the benefits of both.
Also, each working group can be tied to a GSoC project, mentored
by people from the community.
This will add an extra level of seriousness to the D project as I
prefer to call it.
