www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Disallowing the creation of objects using new should have default

reply 12345swordy <alexanderheistermann gmail.com> writes:
It makes no sense otherwise. This logically implies that manual 
memory management is required, yet there is a possibility that 
the parent of the class may use the garbage collection. Which in 
this case, it begs the question on why the GC is forbid in the 
first place.

Does anyone dispute this?
Mar 17 2018
next sibling parent Adam D. Ruppe <destructionator gmail.com> writes:
On Saturday, 17 March 2018 at 13:30:25 UTC, 12345swordy wrote:

 Does anyone dispute this?


I don't really see how this would help anything.
Mar 17 2018
prev sibling parent Basile B. <b2.temp gmx.com> writes:
On Saturday, 17 March 2018 at 13:30:25 UTC, 12345swordy wrote:

 It makes no sense otherwise. This logically implies that manual 
 memory management is required, yet there is a possibility that 
 the parent of the class may use the garbage collection. Which 
 in this case, it begs the question on why the GC is forbid in 
 the first place.

 Does anyone dispute this?


No, it's even required. Mixing GC and non-GC is a known source of 
memory errors, but rather in the opposite way that the one you 
describe. Errors happen when non-GC instances has GC-managed 
members.

To be consistent on this point there are templates like in the 
Containers library (ShouldAddGcRange) or like in IZ 
(MustAddGcRange). The latter is inspired by the first and is 
enhanced with a system of UDA ( NoGc). Also compilation can be 
stopped if by error a GC-managed field is found.

After developing this in IZ, many memory problems i had in KHEOPS 
suddenly disappeared and a kind of animated 2D scene started to 
run for hours without disappearing elements (previously a GC 
collection was causing this).

Enriched by this experience i can confirm that there's nothing to 
dispute at all.
You're 100% correct.
Mar 18 2018