The "no gc" crowd

[Adam D. Ruppe]

Whereas I'm right now with more time than money, I've been 
thinking about doing a kickstarter or something with the goal of 
doing the necessary changes to make going without the gc easy:

1) make sure all allocations in phobos are explicit and avoidable 
where possible
2) have a flag where you can make gc allocations throw an assert 
error at runtime for debugging critical sections
3) maybe the @nogc thing people are talking about

and whatever else comes up.


While I tend to agree that the gc arguments don't carry much 
water (much bigger problem IMO is how intertwined phobos is... 
and even that is meh much of the time), having all this set up 
would be nice slam counter argument: the stdlib does not require 
the gc and there's ways to avoid it pretty easily.


The downside I expect with the kickstarter thing though is most 
the people complaining about this probably aren't willing to 
pitch in cash - they have little to gain from it since they 
aren't D users already. But it'd still be interesting to try.