On 10/16/15 11:03 AM, German Diago wrote:
- Garbage collector. I think there was a plan for Phobos without GC, but...
what about the run-time, can be disabled? I am not sure this meets the
requirements of some embedded devices I work/have worked with.
- Memory-control: Allocators. I saw this has been solved.
- Production-readiness: when I go to C++, the ecosystem is simply
unbeatable.
This keeps me away from moving to D.
- Platform support: For C++, I can use it in phones, embedded, PCs...
basically
everywhere.
What areas are considered "incomplete" as of now to consider D a
production-ready product, in your opinion?
D is production ready for the simple reason it is being factually used
in production by a variety of companies. That said the same definition
does not work for all companies and clearly D does not currently pass
muster to the extent more established languages are.
Reducing the use of GC (both in the standard library and in client code)
is front and center among our priorities. Walter has done a lot of work
on it and we hope others to follow the lead.
For platform support (e.g. mobile) we need the appropriate champions to
take us from "it could be done" and "it experimentally works" all the
way to deliverable tools.
The areas I consider incomplete:
* Language definition, e.g. "shared".
* Language definition _writeup_, we need to be a lot more precise than
we currently are.
* Process for introducing new features, i.e. right now we seem to have
some of the drawbacks of a large political organization and also the
drawbacks of a small community.
* Parts of stdlib, e.g. no robust idioms for transferring complex
objects across threads, unneeded use of the GC, insufficient support for
safe garbage collection; also no extensive containers, file formats,
etc. etc.
* Tutorials - there's no simple tutorial material that takes people from
novice to initiated status.
Andrei
