DIP1028 - Rationale for accepting as is

[Panke]

On Monday, 25 May 2020 at 12:22:25 UTC, Zoadian wrote:
>
> there is no such thing as a trustworthy @trusted. not with how 
> the whole safety system work now.
> you can break previously verified @trusted code by just writing 
> @safe code today.

Do you have an example of code that was correctly labeled 
@trusted in DMD version X, which may not be labeled @trusted in 
DMD version Y because of changes to @safe/@trusted@/@system?