On Monday, 18 July 2016 at 13:48:16 UTC, Andrew Godfrey wrote:
1) As you say, a vision for D3. Maybe just a summary of the
things that are now agreed upon, e.g. autodecoding (though even
there, I think the details of where to move to, are still
contentious. E.g. I personally dislike the convention of "char"
meaning a 1-byte data type but I think some others like it).
2) The case against incremental breaking changes. (I see this
argument somewhat, though it applies less to "dfixable"
breaking changes).
3) Why we feel that breaking changes risk killing D outright.
(I just don't see it. I wonder if we're confusing "dfixable"
breaking changes, with other more disruptive kinds (such as
Tango=>Phobos).)
I wasn't around for the D1 to D2 change, but I was around for
Python 2 to Python 3, which was inconvenient.
My sense is that a lot of the things mentioned here are
"woulda-coulda-shoulda", like having defaults be @safe instead of
@system. Would have been nice to have from the beginning, but
just seems way too disruptive to change it now.
However, I don't have any particular issue with incremental
breaking changes that are dfixable. But I think that saving them
all up to do a huge D3 is potentially more disruptive than doing
a small D3, completely dfixable, then a small D4, etc. Even a D3
that just changed autodecoding (which I don't think is dixable,
but who knows) would be good as it would be just a small limited
breaking change.