Inside D's GC - Hackdot

Re:Inside D's GC by chrisseaton on Tuesday June 20, 2017 @04:19AM (#14592925)

Presumably adding precise stack maps and object maps would be the first step of adding IMMIX. I mean it's the compiler that chooses where to place things in the heap and the stack, so it already knows where they are.

[ Parent | on HN ]

Re:Inside D's GC by readittwice on Tuesday June 20, 2017 @05:02AM (#14593053)

Since, I don't know D, I am curious whether D-code is written in such a way that objects are assumed to be non-moving, especially when interfacing with C.

[ Parent | on HN ]

Re:Inside D's GC by biotronic on Tuesday June 20, 2017 @06:23AM (#14593389)

The spec says the GC may move, though the current GC doesn't do that:

https://dlang.org/spec/garbage.html#obj_pinning_and_gc

There's also been some work on a precise GC for D:

https://forum.dlang.org/post/[email protected]...

[ Parent | on HN ]

Re:Inside D's GC by nicwilson on Tuesday June 20, 2017 @07:11AM (#14593569)

> written in such a way that objects are assumed to be non-moving

Not generally, but you can take the address of anything and shove it into a union or pass it off to C code that stores it somewhere. Because you can, it would be very dangerous to move objects around because of that.

[ Parent | on HN ]

Re:Inside D's GC by chrisseaton on Tuesday June 20, 2017 @05:34AM (#14593202)

Ah, yes maybe FFI and also perhaps inner pointers.

[ Parent | on HN ]