On Wednesday, 20 February 2013 at 18:42:42 UTC, bearophile wrote:
Rob T:
and do a ton of boring stuff like stabilize D2/Phobos before
moving on to D3, install better processes for documentation
and developing the language specifications, and of course
continue to improve the release process (it still needs a real
beta and stable release), etc.
"Stabilize" is the wrong word to use. Implementing the 64
compiler is good, implementing shared libraries is good,
porting D runtime/Phobos to RISC CPUs is good, replacing the GC
is good, improving the floating point management by DMD is
good, and so on and on.
But in my opinion what's more needed now is instead to try to
complete as much as possible the design and implementation of
the missing/broken/incomplete parts of the core language (like
finishing const/immutable design, finishing the implementation
of pure, redesigning properties, fixing @trusted, doing what's
possible with shared, doing what's possible to finish the
inference of tags like pure in templated functions, finishing
the design of packages, finishing the implementation of the
module system, finishing the design of operator overloading,
and so on. The complete list of broken/unfinished parts scares
me).
Bye,
bearophile
Yes your additions is included in what I meant by "stabilize". We
need to freeze the addition of new features for the "stable
release of the language" (we currently do not have a stable
release that is documented) and put it through the acid test of
real world use so it can be polished up based on the feedback.
It may be a good idea to create a team that concentrates only on
releasing a stable version of the language and polishing it up,
otherwise it will never get done. We cannot polish up an
undefined moving target, it's as simple as that.
--rt
