What's the go with the GC these days?

[H. S. Teoh]

On Sun, Jan 06, 2019 at 01:25:32PM -0800, Walter Bright via Digitalmars-d wrote:
> On 1/6/2019 1:40 AM, Manu wrote:
> > It's all irrelevant though, Nobody's asking for multiple pointer
> > types in D. All I want from language support for ARC in D, is an
> > opInc/opDec function which are called appropriately around
> > assignments, and elided appropriately.
> 
> It turns out to be fiendishly difficult to automatically elide counter
> bumps.

Just out of curiosity, any concrete examples of difficulties that
prevent easy elision of counter bumps? Just so we have a better idea of
the challenges we're up against.  Are there any common cases that are
easy enough to implement, that might be "good enough", leaving more
tricky cases aside?


> > copy ctor's can't offer this functionality.
> 
> They can produce a working ref counting solution.

Care to elaborate?  Is it related to how current move + postblit
semantics make it impractical to maintain a consistent ref count?


> D's will have a couple fundamental advantages over the C++ one:
> 
> 1. D won't need the locking on the increment, because D has different
> types for threaded vs non-threaded.
> 
> 2. With dip25 and dip1000, D can offer memory safe access to rc object
> contents.

dip25 and dip1000 have been around for a long time now. What are the
remaining issues blocking their full implementation / deployment?


T

-- 
Give a man a fish, and he eats once. Teach a man to fish, and he will sit forever.