C++17

[Marc Schütz via Digitalmars-d]

On Wednesday, 27 January 2016 at 17:08:09 UTC, Andrei 
Alexandrescu wrote:
> Interesting. It may be worked around, but is arguably a mistake 
> in the language definition. Have you submitted an issue for 
> this? -- Andrei

1) https://issues.dlang.org/show_bug.cgi?id=14246

Includes suggestion on how the compiler could handle it.

2) https://issues.dlang.org/show_bug.cgi?id=9704

Related issue with postblits.

3) https://issues.dlang.org/show_bug.cgi?id=2590

Here, Don suggests that it's not a bug because destructors should 
only run on fully constructed objects, referencing C++. However, 
IMO this is not a valid argument for D, as we have .init, which 
is always a valid state. It can still be debated whether the 
destructor of the failed object itself should run, but it should 
always be safe to run the destructors of members.