On 12/14/16 2:34 PM, ketmar wrote:
On Wednesday, 14 December 2016 at 21:02:09 UTC, Jonathan M Davis wrote:
If we keep making breaking changes, we will never have a significant
user base
so the core issue is a direction D developement should go:
1) have a good language, or
2) have big userbase.
it looks like those goals are in conflict now. while i can see how
userbase matters, i also can't see how it matters for me -- it turns out
that i got a mediocre language as a result. well, there are alot of
"acceptable" languages out there, and c++ has a huge userbase and a huge
codebase, so it will always win here. the only way for D to win (as i
see it) is to deliver a better language. and that means dropping support
for old code from time to time, not to stick with bad design forever.
also, tools like dfix can be made to "upgrade" code.
so far being "stable" didn't brought Bick Bucks or Big Corporate Support
to D. yet instead of using that to advance the language, to redesign
features and so on, D is stuck in a hope of getting some Big Future
Support. i may be completely wrong, of course, but i see the unique
strength that D can exploit: the ability to change.
sure, turning D into "moving target" will make some older code invalid.
but if the author doesn't want to maintain his code, is there any real
reason to use it? with automatic upgrade utility it wouldn't be that
hard to keep the code up-to-date.
i believe that pediodical "cleanups" will make D better, and will win
more users in the long run. so i will continue advocating "moving
target" concept from time to time. ;-)
I've noticed that recent languages like Go and Swift are trying to have
both by bundling a code fixer with new versions of their language. I
have a hard time seeing the downsides of that.
