On Saturday, 14 September 2013 at 06:18:02 UTC, Jonathan M Davis
wrote:
On Friday, September 13, 2013 23:07:03 H. S. Teoh wrote:
OTOH, I find myself switching to classes just to get the
reference
semantics in other cases, even if I never actually do any
inheritance.
Trying to do reference semantics with structs, while certainly
possible,
is just too error-prone IME. Just a few days ago, I
encountered what
looked like a nasty functionality bug in my program, only to
eventually
discover that it was caused by a missing 'ref' in a function's
struct
parameter, so updates to the struct didn't persist as the code
assumed
it would. I found myself seriously considering using classes
instead,
just for the default ref semantics.
In general, if you want to have structs with reference
semantics, it's
probably better to just give them reference semantics by making
it so that any
of their members which are value types are on the heap or by
putting all of
the struct's guts on the heap. At some point, it becomes
debatable as to
whether that's better than using a class, but it does have less
overhead.
There's also RefCounted, which does incur the bookkeeping
overhead of the
refcounting, but it does make it so that the memory is freed as
soon as you
don't need the object anymore.
- Jonathan M Davis
I have a few issues with ref counted.
First, if you *ever* place one in an array or an AA, then you
will leak. It is NOT designed for simply incorporating reference,
but for deterministic finalization.
Second, it has an elaborate postblit and opAssign. This is not a
big issue in itself, but their sole existence does cause dmd to
generate code that is sub-optimal. It also means it can't take
the "optimal" route in a lot of algorithms (array has to emplace
each element individually, for example).
Finally, a RefCounted will implicitly cast to its payload. I
think this is *horrible*. Before you know it, the Payload will
have "jettisoned" its wrapper, and you'll be operating on your
value-type payload "raw". For example:
void foo(T t);
RefCounted!T myRecCounted;
foo(myRefCounted); //Passes. Oops!
The workaround would be to either require explicit "get" to go
from ref counted to payload (breaking existing code), or to
entirelly wrap the RefCounted as a member inside the struct
(requires boilerplate code).
--------
Overall, if I need a reference semantic struct, I find it much
simpler for it to just hold a GC pointer to a payload.
Though to be honest, as H.S. Teoh, I seriously wonder why I even
bother, when I could just use a final class. With proper "private
constructors + non-member "make" function", you can make your
choice outright transparent to the final user too, meaning you
can "fast prototype" with classes, and later change to structs if
you think it is worth it.